FLYNC: a machine-learning-driven framework for discovering long noncoding RNAs in Drosophila melanogaster

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Noncoding RNAs play crucial roles in disease mechanisms.
• The noncoding genome of Drosophila melanogaster, a key disease model organism, remains largely uncharacterized.

Purpose of the Study:

• To develop and validate FLYNC, a novel machine learning model for accurate long noncoding RNA (lncRNA) discovery and classification in Drosophila.
• To identify novel tissue- and cell-specific lncRNAs in Drosophila using FLYNC.

Main Methods:

• Development of FLYNC, an explainable boosting machine model for predicting lncRNA probability.
• Integration of FLYNC into a bioinformatics pipeline for processing RNA sequencing data (single-cell and bulk).
• Application of FLYNC to Drosophila adult head and neural stem cell transcriptomic data.

Main Results:

• FLYNC accurately predicts the likelihood of a transcript being an lncRNA.
• Identification of several novel tissue- and cell-specific lncRNAs in Drosophila.
• Experimental validation of predicted lncRNAs using RT-PCR and RNA PolII binding assays.

Conclusions:

• FLYNC is a robust tool for identifying lncRNAs in Drosophila, overcoming current limitations.
• This work expands the annotation of the Drosophila noncoding genome and provides a powerful machine learning approach for ncRNA discovery.