10-fold cross validation was performed using Support Vector Machine (SVM).
We achieved maximum Pearson Correlation Coefficient (PCC) of 0.801 using SVM between the actual and predicted efficacy values. Similar performance was achieved on Homo-2110.
Workbench is divided into three main modules (I) SMEPred (II) MultiModGen (III)Tool-siMEpred
SMEPred-pipeline generates normal siRNAs and their single modifications and further predict their score against every modified siRNA sequence which indicates its activity i.e how much active it is in silencing the target gene.
MultiModGen-Algorithm helps to generate various permutations and combinations of chemical modifications on different positions as per the users choice.
Tool-siMEpred is provided different models to explore effect of modifiation on different fragments of antisense strand of siRNA.