review article protein-protein interaction detection: methods...

Review ArticleProtein-Protein Interaction Detection Methods and Analysis

V Srinivasa Rao1 K Srinivas1 G N Sujini2 and G N Sunand Kumar1

1 Department of CSE VR Siddhartha Engineering College Vijayawada 520007 India2Department of CSE Mahatma Gandhi Institute of Technology Hyderabad 500075 India

Correspondence should be addressed to V Srinivasa Rao drvsrao9gmailcom

Received 26 October 2013 Revised 5 December 2013 Accepted 20 December 2013 Published 17 February 2014

Academic Editor Yaoqi Zhou

Copyright copy 2014 V Srinivasa Rao et alThis is an open access article distributed under theCreativeCommonsAttribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

Protein-protein interaction plays key role in predicting the protein function of target protein and drug ability of molecules Themajority of genes and proteins realize resulting phenotype functions as a set of interactions The in vitro and in vivomethods likeaffinity purification Y2H (yeast 2 hybrid) TAP (tandem affinity purification) and so forth have their own limitations like cost timeand so forth and the resultant data sets are noisy and have more false positives to annotate the function of drug molecules Thusin silicomethods which include sequence-based approaches structure-based approaches chromosome proximity gene fusion insilico 2 hybrid phylogenetic tree phylogenetic profile and gene expression-based approacheswere developed Elucidation of proteininteraction networks also contributes greatly to the analysis of signal transduction pathways Recent developments have also led tothe construction of networks having all the protein-protein interactions using computational methods for signaling pathways andprotein complex identification in specific diseases

1 Introduction

Protein-protein interactions (PPIs) handle a wide range ofbiological processes including cell-to-cell interactions andmetabolic and developmental control [1] Protein-proteininteraction is becoming one of the major objectives ofsystem biology Noncovalent contacts between the residueside chains are the basis for protein folding protein assemblyand PPI [2] These contacts induce a variety of interactionsand associations among the proteins Based on their con-trasting structural and functional characteristics PPIs can beclassified in several ways [3] On the basis of their interactionsurface they may be homo- or heterooligomeric as judgedby their stability they may be obligate or nonobligate asmeasured by their persistence they may be transient orpermanent [4] A given PPI may be a combination of thesethree specific pairs The transient interactions would formsignaling pathways while permanent interactions will forma stable protein complex

Typically proteins hardly act as isolated species while per-forming their functions in vivo [5] It has been revealed thatover 80 of proteins do not operate alone but in complexes[6] The substantial analysis of authenticated proteins revealsthat the proteins involved in the same cellular processes are

repeatedly found to be interacting with each other [7] Thestudy of PPIs is also important to infer the protein functionwithin the cell The functionality of unidentified proteins canbe predicted on the evidence of their interaction with a pro-tein whose function is already revealed The detailed studyof PPIs has expedited the modeling of functional pathwaysto exemplify the molecular mechanisms of cellular processes[4] Characterizing the interactions of proteins in a givenproteome will be phenomenal to figure out the biochemistryof the cell [4] The result of two or more proteins interactingwith a definite functional objective can be established inseveral ways The significant properties of PPIs have beenmarked by Phizicky and Fields [8]

PPIs can

(i) modify the kinetic properties of enzymes(ii) act as a general mechanism to allow for substrate

channeling(iii) construct a new binding site for small effector mole-

cules(iv) inactivate or suppress a protein(v) change the specificity of a protein for its substrate

through interaction with different binding partners

Hindawi Publishing CorporationInternational Journal of ProteomicsVolume 2014 Article ID 147648 12 pageshttpdxdoiorg1011552014147648

2 International Journal of Proteomics

(vi) serve a regulatory role in either upstream or down-stream level

Uncovering protein-protein interaction informationhelps in the identification of drug targets [9] Studies haveshown that proteins with larger number of interactions(hubs) can include families of enzymes transcription factorsand intrinsically disordered proteins among others [10 11]However PPIs involve more heterogeneous processes andthe scope of their regulation is large For more accurateunderstanding of their importance in the cell one has toidentify various interactions and determine the aftermath ofthe interactions [4]

In recent years PPI data have been enhanced by guar-anteed high-throughput experimental methods such as two-hybrid systems mass spectrometry phage display and pro-tein chip technology [4] Comprehensive PPI networks havebeen built from these experimental resources However thevoluminous nature of PPI data is imposing a challenge to lab-oratory validation Computational analysis of PPI networksis increasingly becoming a mandatory tool to understand thefunctions of unexplored proteins At present protein-proteininteraction (PPI) is one of the key topics for the developmentand progress of modern systemrsquos biology

2 Classification of PPI Detection Methods

Protein-protein interaction detection methods are categor-ically classified into three types namely in vitro in vivoand in silico methods In in vitro techniques a given pro-cedure is performed in a controlled environment outsidea living organism The in vitro methods in PPI detectionare tandem affinity purification affinity chromatographycoimmunoprecipitation protein arrays protein fragmentcomplementation phage display X-ray crystallography andNMR spectroscopy In in vivo techniques a given procedureis performed on the whole living organism itself The in vivomethods in PPI detection are yeast two-hybrid (Y2H Y3H)and synthetic lethality In silico techniques are performedon a computer (or) via computer simulation The in silicomethods in PPI detection are sequence-based approachesstructure-based approaches chromosome proximity genefusion in silico 2 hybrid mirror tree phylogenetic treeand gene expression-based approaches The diagrammaticclassification was given in Table 1

21 In Vitro Techniques to Predict Protein-Protein InteractionsTAP tagging was developed to study PPIs under the intrinsicconditions of the cell [12] Gavin et al first attempted theTAP-tagging method in a high-throughput manner to analyse theyeast interactome [13] This method is based on the doubletagging of the protein of interest on its chromosomal locusfollowed by a two-step purification process [14] Proteinsthat remain associated with the target protein can then beexamined and identified through SDS-PAGE [15] followed bymass spectrometry analysis [15] thereby identifying the PPIcollaborator of the original protein of interest An importantdominance of TAP-tagging is its ability to identify a widevariety of protein complexes and to test the activeness of

monomeric or multimeric protein complexes that exist invivo [14] The TAP when used with mass spectroscopy (MS)will identify protein interactions and protein complexes

The advantage of the affinity chromatography is that itis highly responsive can even detect weakest interactions inproteins and also tests all the sample proteins equally forinteraction with the coupled protein in the column Howeverfalse positive results also arise in the column due to highspecificity among proteins even though they do not getinvolved in the cellular systemThus protein interaction stud-ies cannot fully rely on affinity chromatography and hencerequire other methods in order to crosscheck and verifyresults obtained The affinity chromatography can also beassociated with SDS-PAGE technique andmass spectroscopyin order to generate a high-throughput data

Coimmunoprecipitation confirms interactions using awhole cell extract where proteins are present in their nativeform in a complex mixture of cellular components that maybe required for successful interactions In addition use ofeukaryotic cells enables posttranslational modification whichmay be essential for interaction and which would not occurin prokaryotic expression systems

Protein microarrays are rapidly becoming established asa powerful means to detect proteins monitor their expres-sion levels and probe protein interactions and functionsA protein microarray is a piece of glass on which variousmolecules of protein have been affixed at separate locationsin an ordered manner [16] Protein microarrays have seentremendous progress and interest at the moment and havebecome one of the active areas emerging in biotechnologyThe objective behind protein microarray development isto achieve efficient and sensitive high-throughput proteinanalysis carrying out large numbers of determinations inparallel by automated process

Protein-fragment complementation assay is anothermethod of proteomics for the identification of protein-protein interactions in biological systems Protein-fragmentcomplementation assays (PCAs) are a family of assays fordetecting protein-protein interactions (PPIs) that have beenintroduced to provide simple and direct ways to study PPIs inany living cell multicellular organism or in vitro [17] PCAscan be used to detect PPI between proteins of any molecularweight and expressed at their endogenous levels The twochoices for protein identification using a mass spectroscopyare peptide fingerprinting and shotgun proteomics [18] Forpeptide fingerprinting the eluted complex is separated usingSDS-PAGE The gel is either Coomassie-stained or silver-stained and bands unique to the test sample and hope-fully containing a single protein are excised enzymaticallydigested and analyzed by mass spectrometry The mass ofthese peptides is determined and matched to a peptidedatabase to determine the source protein The gel alsoprovides a rough estimate of the molecular weight of theprotein Since only unique bands are cut out backgroundbands are not identified Abundant background proteins mayobscure target proteins while less abundant proteins may fallbelow the limits of detection by staining This method workswell with purified samples containing only a handful of pro-teins Alternatively for shotgun proteomics the entire eluate

International Journal of Proteomics 3

Table 1 Summary of PPI detection methods

Approach Technique Summary

In vitro

Tandem affinity purification-massspectroscopy (TAP-MS)

TAP-MS is based on the double tagging of the protein of interest onits chromosomal locus followed by a two-step purification processand mass spectroscopic analysis

Affinity chromatographyAffinity chromatography is highly responsive can even detectweakest interactions in proteins and also tests all the sample proteinsequally for interaction

CoimmunoprecipitationCoimmunoprecipitation confirms interactions using a whole cellextract where proteins are present in their native form in a complexmixture of cellular components

Protein microarrays (H) Microarray-based analysis allows the simultaneous analysis ofthousands of parameters within a single experiment

Protein-fragment complementationProtein-fragment complementation assays (PCAs) can be used todetect PPI between proteins of any molecular weight and expressed attheir endogenous levels

Phage display (H) Phage-display approach originated in the incorporation of the proteinand genetic components into a single phage particle

X-ray crystallographyX-ray crystallography enables visualization of protein structures atthe atomic level and enhances the understanding of proteininteraction and function

NMR spectroscopy NMR spectroscopy can even detect weak protein-protein interactions

In vivoYeast 2 hybrid (Y2H) (H) Yeast two-hybrid is typically carried out by screening a protein of

interest against a random library of potential protein partners

Synthetic lethality Synthetic lethality is based on functional interactions rather thanphysical interaction

In silico

Ortholog-based sequence approachOrtholog-based sequence approach based on the homologous natureof the query protein in the annotated protein databases usingpairwise local sequence algorithm

Domain-pairs-based sequenceapproach

Domain-pairs-based approach predicts protein interactions based ondomain-domain interactions

Structure-based approaches Structure-based approaches predict protein-protein interaction if twoproteins have a similar structure (primary secondary or tertiary)

Gene neighborhoodIf the gene neighborhood is conserved across multiple genomes thenthere is a potential possibility of the functional linkage among theproteins encoded by the related genes

Gene fusion

Gene fusion which is often called as Rosetta stone method is basedon the concept that some of the single-domain containing proteins inone organism can fuse to form a multidomain protein in otherorganisms

In silico 2 hybrid (I2H)The I2H method is based on the assumption that interacting proteinsshould undergo coevolution in order to keep the protein functionreliable

Phylogenetic tree The phylogenetic tree method predicts the protein-protein interactionbased on the evolution history of the protein

Phylogenetic profile The phylogenetic profile predicts the interaction between twoproteins if they share the same phylogenetic profile

Gene expression

The gene expression predicts interaction based on the idea thatproteins from the genes belonging to the commonexpression-profiling clusters are more likely to interact with eachother than proteins from the genes belonging to different clusters


containing many proteins is digested Shotgun proteomicsis currently the most powerful strategy for analyzing suchcomplicated mixtures

There are different implementations of the phage displaymethodology as well as different applications [19] One ofthe most common approaches used is the M13 filamentousphage The DNA encoding the protein of interest is ligatedinto the gene encoding one of the coat proteins of the virionNormally the process is followed by computational identi-fication of potential interacting partners and a yeast two-hybrid validation step but themethod is a new born one [20]

X-ray crystallography [21] is essentially a form of veryhigh resolution microscopy which enables visualization ofprotein structures at the atomic level and enhances the under-standing of protein function Specifically it shows how pro-teins interact with other molecules and the conformationalchanges in case of enzymes Armed with this informationwe can also design novel drugs that target a particular targetprotein

In the recent past the researchers have shown interestin the analysis of protein-protein interaction by nuclearmagnetic resonance (NMR) spectroscopy [21] The loca-tion of binding interface is a crucial aspect in the proteininteraction analysis The basis for the NMR spectroscopy isthat magnetically active nuclei oriented by a strong magneticfield absorb electromagnetic radiation at characteristic fre-quencies governed by their chemical environment [22 23]

22 In Vivo Techniques to Predict Protein-Protein InteractionsY2Hmethod is an in vivomethod applied to the detection ofPPIs [24] Two protein domains are required in the Y2H assaywhich will have two specific functions (i) a DNA bindingdomain (DBD) that helps binding to DNA and (ii) an acti-vation domain (AD) responsible for activating transcriptionof DNA Both domains are required for the transcription of areporter gene [25] Y2H analysis allows the direct recognitionof PPI between protein pairsHowever themethodmay incura large number of false positive interactions On the otherhand many true interactions may not be traced using Y2Hassay leading to false negative results In an Y2H assay theinteracting proteins must be localized to the nucleus sinceproteins which are less likely to be present in the nucleusare excluded because of their inability to activate reportergenes Proteins which need posttranslational modificationsto carry out their functions are unlikely to behave or interactnormally in an Y2H experiment Furthermore if the proteinsare not in their natural physiological environment they maynot fold properly to interact [26] During the last decade Y2Hhas been enriched by designing new yeast strains containingmultiple reporter genes and new expression vectors to facil-itate the transformation of yeast cells with hybrid proteins[27] Other widely used techniques such as biolumines-cence resonance energy transfer (BRET) fluorescence reso-nance energy transfers (FRET) and bimolecular fluorescencecomplementation (BiFC) require extensive instrumentationFRET uses time-correlated single-photon counting to predictprotein interactions [28]

Synthetic lethality is an important type of in vivo geneticscreening which tries to understand the mechanisms that

allow phenotypic stability despite genetic variation envi-ronmental changes and random events such as mutationsThis methodology produces mutations or deletions in twoor more genes which are viable alone but cause lethalitywhen combined together under certain conditions [29ndash33] Compared with the results obtained in the aforesaidmethods the relationships detected by synthetic lethalitydo not require necessity of physical interaction between theproteins Therefore we refer to this type of relationships asfunctional interactions

23 In Silico Methods for the Prediction of Protein-ProteinInteractions The yeast two-hybrid (Y2H) system and otherin vitro and in vivo approaches resulted in large-scale devel-opment of useful tools for the detection of protein-proteininteractions (PPIs) between specified proteins that mayoccur in different combinations However the data generatedthrough these approaches may not be reliable because ofnonavailability of possible PPIs In order to understand thetotal context of potential interactions it is better to developapproaches that predict the full range of possible interactionsbetween proteins [4]

A variety of in silico methods have been developed tosupport the interactions that have been detected by exper-imental approach The computational methods for in silicoprediction include sequence-based approaches structure-based approaches chromosome proximity gene fusion insilico 2 hybrid mirror tree phylogenetic tree gene ontologyand gene expression-based approaches The list of all web-servers of in silicomethods was given in Table 2

231 Structure-Based Prediction Approaches The ideabehind the structure-based method is to predict protein-protein interaction if two proteins have a similar structureTherefore if two proteins A and B can interact with eachother then there may be two other proteins A1015840 and B1015840 whosestructures are similar to those of proteins A and B then it isimplied that proteins A1015840 and B1015840 can also interact with eachother Butmost proteinsmay not be having known structuresthe first step for this method is to guess the structure of theprotein based on its sequence This can be done in differentways The PDB database offers useful tools and informationresources for researchers to build the structure for a queryprotein [34] Using the multimeric threading approach Lu etal [35] have made 2865 protein-protein interactions in yeastand 1138 interactions have been confirmed in the DIP [36]

Recently Hosur et al [37] developed a new algorithmto infer protein-protein interactions using structure-basedapproach The Coev2Net algorithm which is a three-stepprocess involves prediction of the binding interface evalu-ation of the compatibility of the interface with an interfacecoevolution based model and evaluation of the confidencescore for the interaction [37] The algorithm when applied tobinary protein interactions has boosted the performance ofthe algorithm over existing methods [38] However Zhang etal [39] have used three-dimensional structural informationto predict PPIs with an accuracy and coverage that aresuperior to predictions based on nonstructural evidence


Table 2 The list of web servers with their references

S number Web server Function Reference

1 Struct2NetThe Struct2Net server makesstructure-based computationalpredictions of protein-proteininteractions (PPIs)

httpgroupscsailmiteducbstruct2netwebserver

2 Coev2NetCoev2Net is a general framework topredict assess and boost confidence inindividual interactions inferred from ahigh-throughput experiment

httpgroupscsailmiteducbcoev2net

3 PRISM PROTOCOLPRISM PROTOCOL is a collection ofprograms that can be used to predictprotein-protein interactions using proteininterfaces

httpprismccbbkuedutrprism protocol

4 InterPreTS InterPreTS uses tertiary structure topredict interactions httpwwwrussellembldeinterprets

5 PrePPIPrePPI predicts protein interactionsusing both structural and nonstructuralinformation

httpbhappc2b2columbiaeduPrePPI

6 iWARPiWARP is a threading-based method topredict protein interaction from proteinsequences

httpgroupscsailmiteducbiwrap

7 PoiNet PoiNet provides PPI filtering and networktopology from different databases httppoinetbioinformaticstw

8 PreSPI PreSPI predicts protein interactions usinga combination of domains httpcodegooglecompprespi

9 PIPE2PIPE2 queries the protein interactionsbetween two proteins based on specificityand sensitivity

httpcgmlabcarletoncaPIPE2

10 HomoMINTHomoMINT predicts interaction inhuman based on ortholog information inmodel organisms

httpmintbiouniroma2itHomoMINT

11 SPPS SPPS searches protein partners of asource protein in other species httpmdlshsmueducnSPPS

12 OrthoMCL-DBOrthoMCL-DB is a graph-clusteringalgorithm designed toidentify homologous proteins based onsequence similarity

httporthomclorgorthomcl

13 P-PODP-POD provides an easy way to find andvisualize the orthologs to a querysequence in the eukaryotes

httpppodprincetonedu

14 COG COG shows phylogenetic classification ofproteins encoded in genomes httpwwwncbinlmnihgovCOG

15 BLASTO BLASTO performs BLAST based onortholog group data httpoxytrichaprincetoneduBlastO

16 PHOG PHOG web server identifies orthologsbased on precomputed phylogenetic trees httpphylogenomicsberkeleyeduphog

17 G-NESTG-NEST is a gene neighborhood scoringtool to identify co-conservedcoexpressed genes

httpsgithubcomdglemayG-NEST

18 InPrePPI InPrePPI predicts protein interactions inprokaryotes based on genomic context httpinpreppibiosinoorgInPrePPIindexjsp

19 STRINGSTRING database includes proteininteractions containing both physical andfunctional associations

httpstringemblde

20 MirrorTreeThe MirrorTree allows graphical andinteractive study of the coevolution oftwo protein families and asseses theirinteractions in a taxonomic context

httpcsbgcnbcsicesmtserver


Table 2 Continued


21 TSEMATSEMA predicts the interaction betweentwo families of proteins based on MonteCarlo approach

httptsemabioinfocnioes

232 Sequence-Based Prediction Approaches Predictions ofPPIs have been carried out by integrating evidence of knowninteractions with information regarding sequential homol-ogyThis approach is based on the concept that an interactionfound in one species can be used to infer the interaction inother species However recently Hosur et al [37] developeda new algorithm to predict protein-protein interactions usingthreading-based approach which takes sequences as inputThe algorithm iWARP (Interface Weighted RAPtor) whichpredicts whether two proteins interact by combining a novellinear programming approach for interface alignment with aboosting classifier [37] for interaction prediction GuilhermeValente et al introduced a new method called Universal InSilico Predictor of Protein-Protein Interactions (UNISPPI)based on primary sequence information for classifying pro-tein pairs as interacting or noninteracting proteins [40]Kernelmethods are hybridmethodswhich use a combinationof properties like protein sequences gene ontologies and soforth [41] However there are two different methods undersequence-based criterion

(1) Ortholog-Based Approach The approach for sequence-based prediction is to transfer annotation from a functionallydefined protein sequence to the target sequence based on thesimilarity Annotation by similarity is based on the homol-ogous nature of the query protein in the annotated proteindatabases using pairwise local sequence algorithm [42]Several proteins from an organism under study may sharesignificant similarities with proteins involved in complexformation in other organisms

The prediction process starts with the comparison of aprobe gene or protein with those annotated proteins in otherspecies If the probe gene or protein has high similarity to thesequence of a gene or proteinwith known function in anotherspecies it is assumed that the probe gene or protein haseither the same function or similar properties Most subunitsof protein complexes were annotated in that way Whenthe function is transferred from a characterized protein toan uncharacterized protein ortholog and paralog conceptsshould be applied Orthologs are the genes in different speciesthat have evolved from a common ancestral gene by specia-tion In contrast paralogs usually refer to the genes related byduplication within a genome [43] In broad sense orthologswill retain the functionality during the course of evolutionwhereas paralogs may acquire new functions Therefore iftwo proteinsmdashA and Bmdashinteract with each other then theorthologs of A and B in a new species are also likely to interactwith each other

(2) Domain-Pairs-Based Approach A domain is a distinctcompact and stable protein structural unit that folds inde-pendently of other such units But most of times domains are

defined as distinct regions of protein sequence that are highlyconserved in the process of evolution As individual struc-tural and functional units protein domains play an importantrole in the development of protein structural class predictionprotein subcellular location prediction membrane proteintype prediction and enzyme class and subclass prediction

Conventionally protein domains are used for basicresearch and also for structure-based drug designing Inaddition domains are directly involved in the intermo-lecular interaction and hence must be fundamental toprotein-protein interactionMultiple studies have shown thatdomain-domain interactions (DDIs) from different exper-iments are more consistent than their corresponding PPIs[44] So it is quite reliable to use the domains and theirinteractions for prediction of the protein-protein interactionsand vice versa [45]

233 Chromosome ProximityGene Neighbourhood Withthe ever increasing number of the completely sequencedgenomes the global context of genes and proteins in thecompleted genomes has provided the researchers with theenriched information needed for the protein-protein interac-tion detection It is well known that the functionally relatedproteins tend to be organized very closely into regions onthe genomes in prokaryotes such as operons the clusters offunctionally related genes transcribed as a single mRNA Ifthe neighborhood relationship is conserved across multiplegenomes then it will bemore relevant for implying the poten-tial possibility of the functional linkage among the proteinsencoded by the related genes And this evidence was appliedto study the functional association of the correspondingproteins This relationship was confirmed by the experimen-tal results and shown to be more independent of relativegene orientation Recently it has been found that thereis functional link among the adjacent bidirectional genesalong the chromosome [46] Interestingly in most casesthe relationship among adjacent bidirectionally transcribedgenes with conserved gene orientation is that one geneencodes a transcriptional regulator and the other belongs tononregulatory protein [47] It has been found that most ofthe regulators control the transcription of the diver gentlytranscribed target geneoperon and automatically regulatetheir own biosynthesis as well This relationship providesanother way to predict the target processes and regulatoryfeatures for transcriptional regulators One of the pitfalls ofthis method is that it is directly suitable for bacterial genomesince gene neighboring is conserved in the bacteria

234 Gene Fusion Gene fusion which is often called asRosetta stone method is based on the concept that some


of the single-domain containing proteins in one organismcan fuse to form a multidomain protein in other organisms[48 49] This domain fusion phenomenon indicates thefunctional association for those separate proteins which arelikely to form a protein complex It has been shown thatfusion events are particularly common in those proteinsparticipating in the metabolic pathway [50 51] This methodcan be used to predict protein-protein interaction by usinginformation of domain arrangements in different genomesHowever it can be applied only to those proteins in whichthe domain arrangement exists

235 In Silico Two-Hybrid (I2h) The method is based onthe assumption that interacting proteins should undergocoevolution in order to keep the protein function reliable Inother words if some of the key amino acids in one proteinchanged the related amino acids in the other protein whichinteracts with the mutated counter partner should also makethe compulsory mutations as well During the analysis phasethe common genomes containing those two proteins willbe identified through multiple sequence alignments and acorrelation coefficient will be calculated for every pair ofresidues in the same protein and between the proteins [52]Accordingly there are three different sets for the pairs twofrom the intraprotein pairs and one from the interproteinpairs The protein-protein interaction is inferred based onthe difference from the distribution of correlation betweenthe interacting partners and the individual proteins SinceI2h analysis is based on the prediction of physical closenessbetween residue pairs of the two individual proteins theresult from this method automatically indicates the possiblephysical interaction between the proteins

236 Phylogenetic Tree Another important method fordetection of interaction between the proteins is phylogenetictree The phylogenetic tree gives the evolution history ofthe protein The mirror tree method predicts protein-proteininteractions under the belief that the interacting proteinsshow similarity in molecular phylogenetic tree because ofthe coevolution through the interaction [53] The underlyingprinciple behind the method is that the coevolution betweenthe interacting proteins can be reflected from the degreeof similarity from the distance matrices of correspondingphylogenetic trees of the interacting proteins [54] The setof organisms common to the two proteins are selected fromthe multiple sequence alignments (MSA) and the results areused to construct the corresponding distance matrix for eachprotein The BLAST scores could also be used to fill thematrices Then the linear correlation is calculated amongthese distance matrices High correlation scores indicate thesimilarity between the phylogenetic trees and therefore theproteins are considered to have the interaction relationshipThe MirrorTree method is used to detect the coevolutionrelationship between proteins and the results are used to inferthe possibility of their physical interaction

237 Phylogenetic Profile The notion for this method is thatthe functionally linked proteins tend to coexist during the

evolution of an organism [55] In other words if two proteinshave a functional linkage in a genome there will be a strongpressure on them to be inherited together during evolutionprocess [51] Thus their corresponding orthologs in othergenome will be preserved or dropped Therefore we candetect the presence or absence (cooccurrence) of proteins inthe phylogenetic profile A phylogenetic profile describes anoccurrence of a certain protein in a set of genomes if twoproteins share the same phylogenetic profiling this indicatesthat the two proteins have the functional linkage In order toconstruct the phylogenetic profile a predetermined thresholdof BLASTP 119864-value is used to detect the presence or absenceof the homological proteins on the target genome with thesource genomes This method gives promising results in thedetection of the functional linkage among the proteins and atthe same time assigns the functions to query proteins Eventhough the phylogenetic profile has shown great potential forbuilding the functional linkage network on the full genomelevel the following two pitfalls should be mentioned oneis that this method is based on full genome sequences andthe other is that the functional linkage between proteins isdetected by their phylogenetic profiling so it is difficult touse the method for those essential proteins in the cell whereno difference can be detected from the phylogenetic profileMoreover even though the increasing number in the sourcegenome set can improve the prediction accuracy there maybe an upper limit for this method

Many genomic events contribute to the noise during thecoevolution such as gene duplication or the possible loss ofgene functions in the course of evolution which could cor-rupt the phylogenetic profile of single genes Phylogenetic-profile-based methods conceded satisfactory performanceonly on prokaryotes but not on eukaryotes [56]

238 Gene Expression The method takes the advantage ofhigh-throughput detection of the whole gene transcriptionlevel in an organism Gene expression means the quantifica-tion of the level at which a particular gene is expressed withina cell tissue or organism under different experimental con-ditions and time intervals By applying the clustering algo-rithms different expression genes can be grouped togetheraccording to their expression levels and the resultant geneexpression under different experimental conditions can helpto enunciate the functional relationships of the various genesA lot of research has also been carried out to investigate therelationship between gene coexpression and protein interac-tion [57] Based on the yeast expression data and proteomedata proteins from the genes belonging to the commonexpression-profiling clusters are more likely to interact witheach other than proteins from the genes belonging to differentclusters In other studies it has been confirmed that adjacentgenes tend to be expressed both in the eukaryotes andprokaryotes The gene coexpression concept is an indirectway to infer the protein interaction suggesting that itmay notbe appropriate for accurate detection of protein interactionsHowever as a complementary approach gene coexpression


can be used to validate interactions generated from otherexperimental methods

3 Comparison of Protein-ProteinInteraction Methods

Each of the above methods has been applied to detectthe protein-protein interaction in both the prokaryotes andeukaryotes The results show that most of them fit betterfor the prokaryotes than eukaryotes [14] The significantincrease for the coverage among those studies during thepast several years could be mainly because of the increasein the number of genomes being decoded This is becausethe more the number of genomes used in the study thehigher the coverage that the methods can reach With theaccumulation of fully sequenced genomes the informationcontent in the reference genome set is expected to increaseAccordingly the prediction accuracy would increase withmore genomes incorporated in the study It can be anticipatedthat withmore andmore genomes available in the future theprediction potential will be improved and the correspondingcombined methods will get higher coverage and accuracyOne thing that should be mentioned is that the selectionof the standard used for the evaluation of the methodshas a great impact on the coverage and accuracy Besidesthe Operon and Swiss-Prot key word recovery used in theabove studies the KEGG has been used as the standard inSearch Tool for the Retrieval of Interacting Genes (STRING)database [58] It can be expected that the prediction coverageand accuracywill be different for eachmethodunder differentstandards Obviously the achieved highest coverage for thegene order method based on the operon standard indicatesthat the method is strongly related to operon

Recent technological advances have allowed high-throughput measurements of protein-protein interactionsin the cell producing protein interaction networks fordifferent species at a rapid pace However high-throughputmethods like yeast two-hybrid MS and phage display haveexperienced high rates of noise and false positives There aresome verification methods to know the reliability of thesehigh throughput interactions They are Expression ProfileReliability (EPR index) Paralogous Verification Method(PVM) [59] Protein Localization Method (PLM) [60] andInteraction Generalities Measures IG1 [61] and IG2 [62] EPR[63] method compares protein interaction with RNA expres-sion profiles whereas PVM analyzes paralogs of interactorsfor comparison The IG1 measure is based on the idea thatinteracting proteins that have no further interactions beyondlevel-1 are likely to be false positivesThe IG2measure uses thetopology of interactions Bayesian approaches have also beenused for calculation of reliability [64ndash66] The PLM gives thetrue positives (TP) as interacting proteins which need to belocalized in the same cellular compartment or annotated tohave a common cellular role So in order to counter theseerrors many methods have been developed which providesconfidence scores with each interaction Also the methodsthat assign scores to individual interactions generallyperform better than those with the set of interactionsobtained from an experiment or a database [67]

4 Computational Analysis of PPI Networks

A PPI network can be described as a heterogeneous networkof proteins joined by interactions as edgesThe computationalanalysis of PPI networks begins with the illustration ofthe PPI network arrangement The simplest sketch takesthe form of a mathematical graph consisting of nodes andedges [68] Protein is represented as a node in such agraph and the proteins that interact with it physically arerepresented as adjacent nodes connected by an edge Anexamination of the network can yield a variety of resultsFor example neighboring proteins in the graph probablymay share more the same functionality In addition to thefunctionality densely connected subgraphs in the networkare likely to form protein complexes as a unit in certainbiological processes Thus the functionality of a proteincan be inferred by spotting at the proteins with which itinteracts and the protein complexes to which it resides Thetopological prediction of new interactions is a novel anduseful option based exclusively on the structural informationprovided by the PPI network (PPIN) topology [69] Somealgorithms like random layout algorithm circular layoutalgorithm hierarchical layout algorithm and so forth areused to visualize the network for further analysis Preciselythe computational analysis of PPI networks is challengingwith these major barriers being commonly confronted [4]

(1) the protein interactions are not stable(2) one protein may have different roles to perform(3) two proteins with distinct functions periodically

interact with each other

5 Role of PPI Networks in Proteomics

Predicting the protein functionality is one of the main objec-tives of the PPI network Despite the recent comprehensivestudies on yeast there are still a number of functionallyunclassified proteins in the yeast database which reflects theimpending need to classify the proteinsThe functional anno-tation of human proteins can provide a strong foundation forthe complete understanding of cell mechanisms informationthat is valuable for drug discovery and development [4] Theincreased availability of PPI networks has developed variouscomputational methods to predict protein functions Theavailability of reliable information on protein interactionsand their presence in physiological and pathophysiologi-cal processes are critical for the development of protein-protein-interaction-based therapeutics The compendium ofall known protein-protein interactions (PPIs) for a given cellor organism is called the interactome

Protein functions may be predicted on the basis ofmodularization algorithms [4] However predictions foundin this way may not be accurate because the accuracy of themodularization process itself is typically low There are othermethods which include the neighbor counting Chi-squareMarkov random field Prodistin and weighted-interactions-based method for the prediction of protein function [77]For greater accuracy protein functions should be predicteddirectly from the topology or connectivity of PPI networks


Table 3 Protein interaction databases

Snumber Database name

Total number ofinteractions(to date)

References Source link Number of speciesorganisms

1 BioGrid 717604 [70] httpthebiogridorg 602 DIP 76570 [36] httpdipdoe-mbiuclaedudipMaincgi 6373 HitPredict 239584 [71] httphintdbhgcjphtp 94 MINT 241458 [72] httpmintbiouniroma2itmint 305 IntAct 433135 [73] httpwwwebiacukintact 86 APID 322579 [74] httpbioinfowdepusalesapidindexhtm 157 BIND gt300000 [75] httpbindca mdash8 PINA20 300155 [76] httpcbggarvanunsweduaupina 7

[4] Several topology-based approaches that predict proteinfunction on the basis of PPI networks have been introducedAt the simplest level the ldquoneighbor counting methodrdquo pre-dicts the function of an unexplored protein by the frequencyof known functions of the immediate neighbor proteins [78]The majority of functions of the immediate neighbors can bestatistically assessed [79] Recently the number of commonneighbors of the known protein and the unknown protein hasbeen taken as the basis for the inference of function [80]Theweighted-graph-mining-based protein function prediction[81] is a novel approach in the area

Protein-protein complex identification is the crucial stepin finding the signal transduction pathways Protein-proteincomplexes mostly consist of antibody-antigen and protease-inhibitor complexes [82] Crystallography is the major toolfor determining protein complexes at atomic resolution

The complete analysis of PPIs can enable better under-standing of cellular organization processes and functionsThe other applications of PPI Network include biologicalindispensability analysis [4] assessing the drug ability ofmolecular targets from network topology [4] estimation ofinteractions reliability [83] identification of domain-domaininteractions [84] prediction of protein interactions [85]detection of proteins involved in disease pathways [86]delineation of frequent interaction network motifs [87]comparison betweenmodel organisms and humans [88] andprotein complex identification [89]

6 Protein Interaction Databases

The massive quantity of experimental PPI data being gener-ated on steady basis has led to the construction of computer-readable biological databases in order to organize and toprocess this data For example the biomolecular interactionnetwork database (BIND) is created on an extensible speci-fication system that permits an elaborate description of themanner in which the PPI data was derived experimentallyoften including links directly to the concluding evidence fromthe literature [75] The database of interacting proteins (DIP)is another database of experimentally determined protein-protein binary interactions [36]Thebiological general repos-itory for interaction datasets (BioGRID) is a database thatcontains protein and genetic interactions among thirteen

different species [70] Interactions are regularly addedthrough exhaustive curation of the primary literature to thedatabases Interaction data is extracted from other databasesincluding BIND and MIPS (Munich Information Center forprotein se quences) [90] as well as directly from large-scaleexperiments [72] HitPredict is a resource of high confidenceprotein-protein interactions from which we can get the totalnumber of interactions in a species for a protein and can viewall the interactions with confidence scores [71]

The Molecular Interaction (MINT) database is anotherdatabase of experimentally derived PPI data extracted fromthe literature with the added element of providing the weightof evidence for each interaction [72] The Human ProteinInteractionDatabase (HPID)was designed to provide humanprotein interaction information precomputed from exist-ing structural and experimental data [91] The informationHyperlinked over Proteins (iHOP) database can be searchedto identify previously reported interactions in PubMed for aprotein of interest [92] IntAct [73] provides an open sourcedatabase and toolkit for the storage presentation and anal-ysis of protein interactions The web interface provides bothtextual and graphical representations of protein interactionsand allows exploring interaction networks in the context ofthe GO annotations of the interacting proteins Howeverwe have observed that the intersection and overlap betweenthese source PPI databases is relatively small Recently theintegration has been done and can be explored in the webserver called APID (Agile Protein Interaction Data Analyzer)which is an interactive bioinformaticsrsquo web tool developed toallow exploration and analysis of currently known informa-tion about protein-protein interactions integrated andunifiedin a common and comparative platform [74] The ProteinInteraction Network Analysis (PINA20) platform is a com-prehensiveweb resource which includes a database of unifiedprotein-protein interaction data integrated from sixmanuallycurated public databases and a set of built-in tools for networkconstruction filtering analysis and visualization [76] Thedatabases and number of interactions were tabled in Table 3

7 Conclusion

While available methods are unable to predict interactionswith 100 accuracy computational methods will scale downthe set of potential interactions to a subset of most likely


interactions These interactions will serve as a starting pointfor further lab experiments The gene expression data andprotein interaction data will improve the confidence ofprotein-protein interactions and the corresponding PPI net-work when used collectively Recent developments have alsoled to the construction of networks having all the protein-protein interactions using computational methods for signaltransduction pathways and protein complex identification inspecific diseases

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgment

Theauthors wish to thank theUniversity Grants Commission(UGC) for extending financial support for this study

References

[1] P Braun and A C Gingras ldquoHistory of protein-protein interac-tions from egg-white to complex networksrdquo Proteomics vol 12no 10 pp 1478ndash1498 2012

[2] Y Ofran and B Rost ldquoAnalysing six types of protein-proteininterfacesrdquo Journal of Molecular Biology vol 325 no 2 pp 377ndash387 2003

[3] I M A Nooren and J M Thornton ldquoDiversity of protein-pro-tein interactionsrdquoThe EMBO Journal vol 22 no 14 pp 3486ndash3492 2003

[4] A Zhang Protein Interaction Networks-Computational Analy-sis Cambridge University Press New York NY USA 2009

[5] M Yanagida ldquoFunctional proteomics current achievementsrdquoJournal of Chromatography B vol 771 no 1-2 pp 89ndash106 2002

[6] T Berggard S Linse and P James ldquoMethods for the detectionand analysis of protein-protein interactionsrdquo Proteomics vol 7no 16 pp 2833ndash2842 2007

[7] C von Mering R Krause B Snel et al ldquoComparative assess-ment of large-scale data sets of protein-protein interactionsrdquoNature vol 417 no 6887 pp 399ndash403 2002

[8] E M Phizicky and S Fields ldquoProtein-protein interactionsmethods for detection and analysisrdquo Microbiological Reviewsvol 59 no 1 pp 94ndash123 1995

[9] C S Pedamallu and J Posfai ldquoOpen source tool for predictionof genome wide protein-protein interaction network based onortholog informationrdquo Source Code for Biology and Medicinevol 5 article 8 2010

[10] A K Dunker M S Cortese P Romero L M Iakoucheva andV N Uversky ldquoFlexible nets the roles of intrinsic disorder inprotein interaction networksrdquo FEBS Journal vol 272 no 20 pp5129ndash5148 2005

[11] M Sarmady W Dampier and A Tozeren ldquoHIV proteinsequence hotspots for crosstalk with host hub proteinsrdquo PLoSONE vol 6 no 8 Article ID e23293 2011

[12] G Rigaut A Shevchenko B Rutz M Wilm M Mann and BSeraphin ldquoA generic protein purification method for proteincomplex characterization and proteome explorationrdquo NatureBiotechnology vol 17 no 10 pp 1030ndash1032 1999

[13] A-C Gavin M Bosche R Krause et al ldquoFunctional organi-zation of the yeast proteome by systematic analysis of proteincomplexesrdquo Nature vol 415 no 6868 pp 141ndash147 2002

[14] S Pitre M Alamgir J R Green M Dumontier F Dehneand A Golshani ldquoComputational methods for predicting pro-tein-protein interactionsrdquo Advances in Biochemical Engineer-ingBiotechnology vol 110 pp 247ndash267 2008

[15] J S Rohila M Chen R Cerny and M E Fromm ldquoImprovedtandem affinity purification tag and methods for isolation ofprotein heterocomplexes from plantsrdquo Plant Journal vol 38 no1 pp S172ndashS181 2004

[16] G MacBeath and S L Schreiber ldquoPrinting proteins as microar-rays for high-throughput function determinationrdquo Science vol289 no 5485 pp 1760ndash1763 2000

[17] S W Michnick P H Ear C Landry M K Malleshaiah and VMessier ldquoProtein-fragment complementation assays for large-scale analysis functional dissection and dynamic studies ofprotein-protein interactions in living cellsrdquo Methods in Molec-ular Biology vol 756 pp 395ndash425 2011

[18] J J Moresco P C Carvalho and J R Yates III ldquoIdentifyingcomponents of protein complexes in C elegans using co-immu-noprecipitation and mass spectrometryrdquo Journal of Proteomicsvol 73 no 11 pp 2198ndash2204 2010

[19] G P Smith ldquoFilamentous fusion phage novel expression vec-tors that display cloned antigens on the virion surfacerdquo Sciencevol 228 no 4705 pp 1315ndash1317 1985

[20] A H Y Tong B Drees G Nardelli et al ldquoA combined experi-mental and computational strategy to define protein interactionnetworks for peptide recognitionmodulesrdquo Science vol 295 no5553 pp 321ndash324 2002

[21] A H Y Tong M Evangelista A B Parsons et al ldquoSystematicgenetic analysis with ordered arrays of yeast deletion mutantsrdquoScience vol 294 no 5550 pp 2364ndash2368 2001

[22] M ROrsquoConnell R Gamsjaeger and J PMackay ldquoThe structur-al analysis of protein-protein interactions by NMR spectrosco-pyrdquo Proteomics vol 9 no 23 pp 5224ndash5232 2009

[23] G Gao J G Williams and S L Campbell ldquoProtein-proteininteraction analysis by nuclear magnetic resonance spec-troscopyrdquo Methods in Molecular Biology vol 261 pp 79ndash922004

[24] P Uetz L Glot G Cagney et al ldquoA comprehensive analysisof protein-protein interactions in Saccharomyces cerevisiaerdquoNature vol 403 no 6770 pp 623ndash627 2000

[25] T Ito T Chiba R Ozawa M Yoshida M Hattori and YSakaki ldquoA comprehensive two-hybrid analysis to explore theyeast protein interactomerdquo Proceedings of the National Academyof Sciences of the United States of America vol 98 no 8 pp4569ndash4574 2001

[26] J I Semple C M Sanderson and R D Campbell ldquoThe jury isout on ldquoguilt by associationrdquo trialsrdquo Brief Funct Genomic Prote-omic vol 1 no 1 pp 40ndash52 2002

[27] P James J Halladay and E A Craig ldquoGenomic libraries and ahost strain designed for highly efficient two-hybrid selection inyeastrdquo Genetics vol 144 no 4 pp 1425ndash1436 1996

[28] D Lleres S Swift andA I Lamond ldquoDetecting protein-proteininteractions in vivo with FRET using multiphoton fluorescencelifetime imaging microscopy (FLIM)rdquo Current Protocols inCytometry vol 12 Unit1210 2007

[29] S L Rutherford ldquoFrom genotype to phenotype bufferingmechanisms and the storage of genetic informationrdquo BioEssaysvol 22 no 12 pp 1095ndash1105 2000


[30] J L Hartman IV B Garvik and L Hartwell ldquoCell biology prin-ciples for the buffering of genetic variationrdquo Science vol 291 no5506 pp 1001ndash1004 2001

[31] A Bender and J R Pringle ldquoUse of a screen for syntheticlethal and multicopy suppressee mutants to identify two newgenes involved in morphogenesis in Saccharomyces cerevisiaerdquoMolecular and Cellular Biology vol 11 no 3 pp 1295ndash1305 1991

[32] S L Ooi X Pan B D Peyser et al ldquoGlobal synthetic-lethalityanalysis and yeast functional profilingrdquo Trends in Genetics vol22 no 1 pp 56ndash63 2006

[33] J A Brown G Sherlock C L Myers et al ldquoGlobal analysisof gene function in yeast by quantitative phenotypic profilingrdquoMolecular Systems Biology vol 2 Article ID 20060001 2006

[34] H Berman K Henrick H Nakamura and J L Markley ldquoTheworldwide Protein Data Bank (wwPDB) ensuring a singleuniform archive of PDB datardquo Nucleic Acids Research vol 35no 1 pp D301ndashD303 2007

[35] L Lu H Lu and J Skolnick ldquoMultiprospector an algorithmfor the prediction of protein-protein interactions bymultimericthreadingrdquo Proteins vol 49 no 3 pp 350ndash364 2002

[36] I Xenarios Ł Salwınski X J Duan P Higney S-M Kimand D Eisenberg ldquoDIP the database of interacting proteins aresearch tool for studying cellular networks of protein interac-tionsrdquo Nucleic Acids Research vol 30 no 1 pp 303ndash305 2002

[37] R Hosur J Xu J Bienkowska and B Berger ldquoIWRAP aninterface threading approach with application to prediction ofcancer-related protein-protein interactionsrdquo Journal of Molecu-lar Biology vol 405 no 5 pp 1295ndash1310 2011

[38] R Hosur J Peng A Vinayagam U Stelzl et al ldquoA computa-tional framework for boosting confidence in high-throughputprotein-protein interaction datasetsrdquo Genome Biology vol 13no 8 p R76 2012

[39] Q C Zhang D Petrey L Deng et al ldquoStructure-based predic-tion of protein-protein interactions on a genome-wide scalerdquoNature vol 490 no 7421 pp 556ndash560 2012

[40] G T Valente M L Acencio C Martins and N Lemke ldquoThedevelopment of a universal in silico predictor of protein-proteininteractionsrdquo PLoS ONE vol 8 no 5 Article ID e65587 2013

[41] A Ben-Hur and W S Noble ldquoKernel methods for predictingprotein-protein interactionsrdquo Bioinformatics vol 21 no 1 ppi38ndashi46 2005

[42] S-A Lee C-H Chan C-H Tsai et al ldquoOrtholog-based pro-tein-protein interaction prediction and its application to inter-species interactionsrdquoBMCBioinformatics vol 9 supplement 12article S11 2008

[43] R L Tatusov E V Koonin and D J Lipman ldquoA genomic per-spective on protein familiesrdquo Science vol 278 no 5338 pp 631ndash637 1997

[44] V Memisevic A Wallqvist and J Reifman ldquoReconstitut-ing protein interaction networks using parameter-dependentdomain-domain interactionsrdquo BMC Bioinformatics vol 14 pp154ndash168 2013

[45] J Wojcik and V Schachter ldquoProtein-protein interaction mapinference using interacting domain profile pairsrdquo Bioinformat-ics vol 17 supplement 1 pp S296ndashS305 2001

[46] M Yamada M S Kabir and R Tsunedomi ldquoDivergent pro-moter organization may be a preferred structure for genecontrol in Escherichia colirdquo Journal of Molecular Microbiologyand Biotechnology vol 6 no 3-4 pp 206ndash210 2003

[47] J Raes J O Korbel M J Lercher et al ldquoPrediction of effectivegenome size in meta genomic samplesrdquo Genome Biology vol 7pp 911ndash917 2004

[48] A J Enright I Illopoulos N C Kyrpides and C A OuzounisldquoProtein interaction maps forcomplete genomes based on genefusion eventsrdquo Nature vol 402 no 6757 pp 86ndash90 1999

[49] EMMarcotte M Pellegrini H-L Ng DW Rice T O Yeatesand D Eisenberg ldquoDetecting protein function and protein-protein interactions from genome sequencesrdquo Science vol 285no 5428 pp 751ndash753 1999

[50] R Overbeek M Fonstein M DrsquoSouza G D Push and NMaltsev ldquoThe use of gene clusters to infer functional couplingrdquoProceedings of the National Academy of Sciences of the UnitedStates of America vol 96 no 6 pp 2896ndash2901 1999

[51] C Freiberg ldquoNovel computational methods in anti-microbialtarget identificationrdquo Drug Discovery Today vol 6 no 2 ppS72ndashS80 2001

[52] F Pazos and A Valencia ldquoIn silico two-hybrid system for theselection of physically interacting protein pairsrdquo Proteins vol47 no 2 pp 219ndash227 2002

[53] T Sato Y Yamanishi M Kanehisa K Horimoto and H TohldquoImprovement of the mirrortree method by extracting evolu-tionary informationrdquo in Sequence andGenomeAnalysisMethodand Applications pp 129ndash139 Concept Press 2011

[54] R A Craig and L Liao ldquoPhylogenetic tree information aidssupervised learning for predicting protein-protein interactionbased on distance matricesrdquo BMC Bioinformatics vol 8 article6 2007

[55] K Srinivas A A Rao G R Sridhar and S Gedela ldquoMethod-ology for phylogenetic tree constructionrdquo Journal of Proteomicsamp Bioinformatics vol 1 pp S005ndashS011 2008

[56] T W Lin J W Wu and D Tien-Hao Chang ldquoCombiningphylogenetic profiling-based andmachine learning-based tech-niques to predict functional related proteinsrdquo PLoS ONE vol 8no 9 Article ID e75940 2013

[57] A Grigoriev ldquoA relationship between gene expression andprotein interactions on the proteome scale analysis of thebacteriophage T7 and the yeast Saccharomyces cerevisiaerdquoNucleic Acids Research vol 29 no 17 pp 3513ndash3519 2001

[58] C von Mering L J Jensen B Snel et al ldquoSTRING known andpredicted protein-protein associations integrated and trans-ferred across organismsrdquo Nucleic Acids Research vol 33 ppD433ndashD437 2005

[59] C M Deane Ł Salwinski I Xenarios and D EisenbergldquoProtein interactions two methods for assessment of the reli-ability of high throughput observationsrdquo Molecular amp CellularProteomics vol 1 no 5 pp 349ndash356 2002

[60] E Sprinzak S Sattath andHMargalit ldquoHow reliable are exper-imental protein-protein interaction datardquo Journal of MolecularBiology vol 327 no 5 pp 919ndash923 2003

[61] R Saito H Suzuki and Y Hayashizaki ldquoConstruction ofreliable protein-protein interaction networks with a new inter-action generality measurerdquo Bioinformatics vol 19 no 6 pp756ndash763 2003

[62] R Saito H Suzuki and Y Hayashizaki ldquoInteraction generalitya measurement to assess the reliability of a protein-proteininteractionrdquoNucleic Acids Research vol 30 no 5 pp 1163ndash11682002

[63] A Mora and I M Donaldson ldquoEffects of protein interactiondata integration representation and reliability on the use ofnetwork properties for drug target predictionrdquo BMC Bioinfor-matics vol 13 p 294 2012

[64] A M Edwards B Kus R Jansen D Greenbaum J Greenblattand M Gerstein ldquoBridging structural biology and genomics


assessing protein interaction data with known complexesrdquoTrends in Genetics vol 18 no 10 pp 529ndash536 2002

[65] R Jansen H Yu D Greenbaum et al ldquoA bayesian networksapproach for predicting protein-protein interactions fromgenomic datardquo Science vol 302 no 5644 pp 449ndash453 2003

[66] S Asthana O D King F D Gibbons and F P Roth ldquoPredict-ing protein complex membership using probabilistic networkreliabilityrdquo Genome Research vol 14 no 6 pp 1170ndash1175 2004

[67] S Suthram T Shlomi E Ruppin R Sharan and T Ideker ldquoAdirect comparison of protein interaction confidence assignmentschemesrdquo BMC Bioinformatics vol 7 article 360 2006

[68] A Wagner ldquoHow the global structure of protein interactionnetworks evolvesrdquo Proceedings of the Royal Society B vol 270no 1514 pp 457ndash466 2003

[69] C Vittorio Cannistraci G Alanis-Lobato and T RavasildquoMinimum curvilinearity to enhance topological prediction ofprotein interactions by network embeddingrdquo Bioinformaticsvol 29 pp i199ndashi209 2013

[70] C Stark B-J Breitkreutz T Reguly L Boucher A Breitkreutzand M Tyers ldquoBioGRID a general repository for interactiondatasetsrdquo Nucleic Acids Research vol 34 pp D535ndash539 2006

[71] A Patil K Nakai and H Nakamura ldquoHitPredict a databaseof quality assessed protein-protein interactions in nine speciesrdquoNucleic Acids Research vol 39 no 1 pp D744ndashD749 2011

[72] A Chatr-aryamontri A Ceol L M Palazzi et al ldquoMINT theMolecular INTeraction databaserdquo Nucleic Acids Research vol35 no 1 pp D572ndashD574 2007

[73] H Hermjakob L Montecchi-Palazzi C Lewington et alldquoIntAct an open source molecular interaction databaserdquoNucleic Acids Research vol 32 pp D452ndashD455 2004

[74] C Prieto and J de Las Rivas ldquoAPID agile protein interactionDataAnalyzerrdquo Nucleic Acids Research vol 34 pp W298ndashW302 2006

[75] G D Bader I Donaldson CWolting B F F Ouellette T Paw-son and CW V Hogue ldquoBINDmdashthe biomolecular interactionnetwork databaserdquoNucleic Acids Research vol 29 no 1 pp 242ndash245 2001

[76] M J Cowley M Pinese K S Kassahn et al ldquoPINA v2 0mining interactome modulesrdquo Nucleic Acids Research vol 40pp D862ndashD865 2012

[77] K S Ahmed and S M El-Metwally ldquoProtein function pre-diction based on protein-protein interactions a comparativestudyrdquo IFMBE Proceedings vol 41 pp 1245ndash1249 2014

[78] B Schwikowski P Uetz and S Fields ldquoA network of protein-protein interactions in yeastrdquo Nature Biotechnology vol 18 no12 pp 1257ndash1261 2000

[79] H Hishigaki K Nakai T Ono A Tanigami and T TakagildquoAssessment of prediction accuracy of protein function fromprotein-protein interaction datardquo Yeast vol 18 no 6 pp 523ndash531 2001

[80] C Lin D Jiang and A Zhang ldquoPrediction of protein func-tion using common-neighbors in protein-protein interactionnetworksrdquo in Proceedings of the 6th IEEE Symposium on BioIn-formatics and BioEngineering (BIBE rsquo06) pp 251ndash260 October2006

[81] T Kim M Li K Ho Ryu and J Shin ldquoPrediction of proteinfunction fromprotein-protein interaction network byWeightedGraph Miningrdquo in Proceedings of the 4th International Confer-ence on Bioinformatics amp BioMedical Technology (IPCBEE rsquo12)vol 29 pp 150ndash154 2012

[82] R C Liddington ldquoStructural basis of protein-protein interac-tionsrdquoMethods in Molecular Biology vol 261 pp 3ndash14 2004

[83] J S Bader A Chaudhuri JM Rothberg and J Chant ldquoGainingconfidence in high-throughput protein interaction networksrdquoNature Biotechnology vol 22 no 1 pp 78ndash85 2004

[84] K S Guimaraes R Jothi E Zotenko and T M PrzytyckaldquoPredicting domain-domain interactions using a parsimonyapproachrdquo Genome Biology vol 7 no 11 article R104 2006

[85] B Lehner and A G Fraser ldquoA first-draft human protein-interaction maprdquo Genome Biology vol 5 no 9 p R63 2004

[86] D R Rhodes S A Tomlins S Varambally et al ldquoProbabilisticmodel of the human protein-protein interaction networkrdquoNature Biotechnology vol 23 no 8 pp 951ndash959 2005

[87] RMilo S Itzkovitz N Kashtan et al ldquoSuperfamilies of evolvedand designed networksrdquo Science vol 303 no 5663 pp 1538ndash1542 2004

[88] T K B Gandhi J Zhong S Mathivanan et al ldquoAnalysis of thehuman protein interactome and comparison with yeast wormand fly interaction datasetsrdquo Nature Genetics vol 38 no 3 pp285ndash293 2006

[89] S Srihari and H waileong ldquoSurvey of computational methodsfor protein complex prediction from protein interaction net-worksrdquo Journal of Bioinformatics and Computational Biologyvol 11 no 2 Article ID 1230002 2013

[90] HWMewes A Ruepp FTheis et al ldquoMIPS curated databasesand comprehensive secondary data resources in 2010rdquo NucleicAcids Research vol 39 no 1 pp D220ndashD224 2011

[91] KHan B ParkH Kim J Hong and J Park ldquoHPID the humanprotein interactionrdquo Bioinformatics vol 20 no 15 pp 2466ndash2470 2004

[92] J M Fernandez R Hoffmann and A Valencia ldquoiHOP webservicesrdquo Nucleic Acids Research vol 35 pp W21ndashW26 2007

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Anatomy Research International

PeptidesInternational Journal of


Hindawi Publishing Corporation httpwwwhindawicom

International Journal of

Volume 2014

Zoology


Molecular Biology International

GenomicsInternational Journal of


The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014


BioinformaticsAdvances in

Marine BiologyJournal of



Signal TransductionJournal of


BioMed Research International

Evolutionary BiologyInternational Journal of



Biochemistry Research International

ArchaeaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Genetics Research International


Advances in

Virolog y

Hindawi Publishing Corporationhttpwwwhindawicom

Nucleic AcidsJournal of

Volume 2014

Stem CellsInternational



Enzyme Research



Microbiology


(vi) serve a regulatory role in either upstream or down-stream level

Uncovering protein-protein interaction informationhelps in the identification of drug targets [9] Studies haveshown that proteins with larger number of interactions(hubs) can include families of enzymes transcription factorsand intrinsically disordered proteins among others [10 11]However PPIs involve more heterogeneous processes andthe scope of their regulation is large For more accurateunderstanding of their importance in the cell one has toidentify various interactions and determine the aftermath ofthe interactions [4]

In recent years PPI data have been enhanced by guar-anteed high-throughput experimental methods such as two-hybrid systems mass spectrometry phage display and pro-tein chip technology [4] Comprehensive PPI networks havebeen built from these experimental resources However thevoluminous nature of PPI data is imposing a challenge to lab-oratory validation Computational analysis of PPI networksis increasingly becoming a mandatory tool to understand thefunctions of unexplored proteins At present protein-proteininteraction (PPI) is one of the key topics for the developmentand progress of modern systemrsquos biology

2 Classification of PPI Detection Methods

Protein-protein interaction detection methods are categor-ically classified into three types namely in vitro in vivoand in silico methods In in vitro techniques a given pro-cedure is performed in a controlled environment outsidea living organism The in vitro methods in PPI detectionare tandem affinity purification affinity chromatographycoimmunoprecipitation protein arrays protein fragmentcomplementation phage display X-ray crystallography andNMR spectroscopy In in vivo techniques a given procedureis performed on the whole living organism itself The in vivomethods in PPI detection are yeast two-hybrid (Y2H Y3H)and synthetic lethality In silico techniques are performedon a computer (or) via computer simulation The in silicomethods in PPI detection are sequence-based approachesstructure-based approaches chromosome proximity genefusion in silico 2 hybrid mirror tree phylogenetic treeand gene expression-based approaches The diagrammaticclassification was given in Table 1

21 In Vitro Techniques to Predict Protein-Protein InteractionsTAP tagging was developed to study PPIs under the intrinsicconditions of the cell [12] Gavin et al first attempted theTAP-tagging method in a high-throughput manner to analyse theyeast interactome [13] This method is based on the doubletagging of the protein of interest on its chromosomal locusfollowed by a two-step purification process [14] Proteinsthat remain associated with the target protein can then beexamined and identified through SDS-PAGE [15] followed bymass spectrometry analysis [15] thereby identifying the PPIcollaborator of the original protein of interest An importantdominance of TAP-tagging is its ability to identify a widevariety of protein complexes and to test the activeness of

monomeric or multimeric protein complexes that exist invivo [14] The TAP when used with mass spectroscopy (MS)will identify protein interactions and protein complexes

The advantage of the affinity chromatography is that itis highly responsive can even detect weakest interactions inproteins and also tests all the sample proteins equally forinteraction with the coupled protein in the column Howeverfalse positive results also arise in the column due to highspecificity among proteins even though they do not getinvolved in the cellular systemThus protein interaction stud-ies cannot fully rely on affinity chromatography and hencerequire other methods in order to crosscheck and verifyresults obtained The affinity chromatography can also beassociated with SDS-PAGE technique andmass spectroscopyin order to generate a high-throughput data

Coimmunoprecipitation confirms interactions using awhole cell extract where proteins are present in their nativeform in a complex mixture of cellular components that maybe required for successful interactions In addition use ofeukaryotic cells enables posttranslational modification whichmay be essential for interaction and which would not occurin prokaryotic expression systems

Protein microarrays are rapidly becoming established asa powerful means to detect proteins monitor their expres-sion levels and probe protein interactions and functionsA protein microarray is a piece of glass on which variousmolecules of protein have been affixed at separate locationsin an ordered manner [16] Protein microarrays have seentremendous progress and interest at the moment and havebecome one of the active areas emerging in biotechnologyThe objective behind protein microarray development isto achieve efficient and sensitive high-throughput proteinanalysis carrying out large numbers of determinations inparallel by automated process

Protein-fragment complementation assay is anothermethod of proteomics for the identification of protein-protein interactions in biological systems Protein-fragmentcomplementation assays (PCAs) are a family of assays fordetecting protein-protein interactions (PPIs) that have beenintroduced to provide simple and direct ways to study PPIs inany living cell multicellular organism or in vitro [17] PCAscan be used to detect PPI between proteins of any molecularweight and expressed at their endogenous levels The twochoices for protein identification using a mass spectroscopyare peptide fingerprinting and shotgun proteomics [18] Forpeptide fingerprinting the eluted complex is separated usingSDS-PAGE The gel is either Coomassie-stained or silver-stained and bands unique to the test sample and hope-fully containing a single protein are excised enzymaticallydigested and analyzed by mass spectrometry The mass ofthese peptides is determined and matched to a peptidedatabase to determine the source protein The gel alsoprovides a rough estimate of the molecular weight of theprotein Since only unique bands are cut out backgroundbands are not identified Abundant background proteins mayobscure target proteins while less abundant proteins may fallbelow the limits of detection by staining This method workswell with purified samples containing only a handful of pro-teins Alternatively for shotgun proteomics the entire eluate




In vitro













In silico






Gene fusion





Gene expression














































httpstringemblde




Table 2 Continued













































7 Conclusion






Acknowledgment


References







































































































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology




In vitro













In silico






Gene fusion





Gene expression














































httpstringemblde




Table 2 Continued













































7 Conclusion






Acknowledgment


References







































































































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology













































httpstringemblde




Table 2 Continued













































7 Conclusion






Acknowledgment


References







































































































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology


Table 2 Continued













































7 Conclusion






Acknowledgment


References







































































































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology


































7 Conclusion






Acknowledgment


References







































































































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology





Acknowledgment


References







































































































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology










































































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology

review article protein-protein interaction detection: methods...

Documents