TAGPT: A Web Server for Prediction of Trait Associated Genes using Gene Expression Data

Abstract

Background: Transcriptomics plays a vital role in functional annotation of genes. This is the well proven fact that genes related to a particular trait can be predicted through modelling and analysis of transcriptomic data. Massive gene expression data related to different useful traits are available in public repositories. However, this valuable information are underutilized due to the lack of appropriate methodologies and efficient tools for identification of related genes and transcription factors. In order to properly utilize this genomic information an efficient algorithm and a sophisticated computational tool based on efficient algorithm is of urgent need. Results: In this study, an efficient algorithm has been developed by utilizing non-linear penalized regression technique Kernelized Least Absolute Shrinkage and Selection Operator (LASSO) for predicting genes related to useful traits using gene expression data. Also, a user friendly Web based tool (TAGPT, Trait Associated Genes Prediction Tool) has been developed based on this algorithm. The efficiency of TAGPT over existing tools has been demonstrated by predicting the trait specific genes using five sets of gene expression data from various experiments related to gene expression. It has been observed that the developed algorithm behind TAGPT is accurate and fast while predicting the specific genes related to a particular trait by analysing the predicted genes. Conclusion: TAGPT is user friendly Web server which provides the list of predicted genes related to a specific trait with their expression value along with their evaluation measures (sensitivity, specificity, classification accuracy etc.). This web server can be accessed by using any web browsers like Internet Explorer, Google Chrome, Mozilla Firefox etc. TAGPT is freely available at http:// cabgrid.res.in/tagpt.