Team Presents Benchmark Study of RNA Classification Tools

Investigators at India's National Agri-Food Biotechnology Institute consider computational strategies for classifying protein coding- and non-coding RNA, using the top tools to put together an RNAChallenge validation test set containing well-supported transcripts. For the analysis, the team used data from 135 transcriptomic datasets to test 24 tools and dozens of computational models for distinguishing coding and non-coding transcripts, ultimately using insights from 48 models contributing to 17 of the computational tools to put together a validation set dubbed RNAChallenge. "This [RNAChallenge] validation set is also helpful in detecting false positives and false negatives, a problem commonly found in the biological domain," the authors explain, noting that their broader analysis of existing computational models pointed to "significant room for improving the tools by addressing these issues."