FEELnc

FEELnc annotates long non-coding RNAs (lncRNAs) from RNA-seq assembled transcripts using an alignment-free Random Forest classifier to distinguish lncRNAs from mRNAs.

Key Features:

• Alignment-Free Annotation: Performs classification without sequence alignment using features such as multi k-mer frequencies and relaxed open reading frames (ORFs).
• Machine Learning Classifier: Implements a Random Forest model trained on general transcript features to separate lncRNAs and mRNAs.
• Benchmarking Performance: Shows similar or superior classification performance compared to five state-of-the-art tools on datasets including GENCODE and NONCODE.
• Customizable Modules: Provides modules to fine-tune classification parameters and formalize lncRNA annotations.
• Training Set Independence: Can identify lncRNAs when no prior annotated non-coding training set is available.

Scientific Applications:

• Canine genome annotation (LUPA consortium): Applied to 20 canine RNA-seq samples, identifying 10,374 novel lncRNAs and 58,640 mRNA transcripts to expand canine genome annotations.

Methodology:

Random Forest models trained on multi k-mer frequencies and relaxed ORFs classify assembled transcripts in an alignment-free manner.