INFLAMeR: a machine learning algorithm based on large-scale perturbation screening identified new lncRNAs regulating differentiation and survival of leukaemia cells

Abstract (English)

Long non-coding RNAs (lncRNAs) are a diverse group of transcripts with poorly understood
functionality. To address this gap, we developed INFLAMeR, an advanced machine learning
model trained on CRISPRi screening data, to predict functional lncRNAs using comprehensive
genetic features. We experimentally validated the predictions by assessing their impact on cell
proliferation and anticancer drug resistance. Among the selected lncRNAs, 85% showed
significant effects upon knockdown, while low-scoring lncRNAs had no discernible impact.
Notably, our study elucidated the functional role of SNHG6 in hematopoietic differentiation.
INFLAMeR greatly enhances the prediction of functional lncRNAs, providing valuable insights
into their regulatory landscape. By integrating INFLAMeR with experimental validation, we can
identify and characterize functional lncRNAs in a cell-type-specific manner, contributing to a
deeper understanding of their involvement in cellular processes. Our findings revealed insights
into lncRNA biology and a framework for improving the identification of functional lncRNAs.