An Integrative System For Prediction Of Nac Proteins In Rice Using Different Feature Extraction Methods

TheNAC gene family encodesa large family of plant-specific transcriptionfactorswithdiverseroles invarious developmental processes and stress responsesin plants.Creationof genome wide prediction toolsfor NAC proteinswillhave a significant impacton gene annotationin rice.In the present study,NACSVM,a tool forcomputationalgenome-scale prediction of NACproteins in ricewasdeveloped integratingcompositional and evolutionary information ofNACproteins.Initially, support vector machine (SVM)-based modules weredeveloped usingcombinatorialpresence of diverse protein featuressuch astraditional amino acid, dipeptide (i+1), tripeptide (i+2),four-partscomposition andPSSMand an overallaccuracy of79%, 93%, 93%, 79%and 100% respectively was achieved.Later,twohybrid modules weredeveloped based on amino acid, dipeptideand tripeptidecomposition,through which anoverall accuracyof83% and 79%was achieved.NACSVM wasalso evaluatedusing position-specific iterated–basic localalignment search toolwhich resulted in a loweraccuracy of 50%.In order to benchmark NACSVM,thetool wasevaluated using independent data test and cross validation methods.The different statisticalanalysescarriedout revealed that the proposed algorithm isan useful toolfor annotatingNAC proteinsingenome of rice