KATMAP infers splicing factor activity and regulatory targets from knockdown data

Area of Science:

• Molecular Biology
• Bioinformatics
• Genomics

Background:

• RNA sequencing (RNA-seq) reveals numerous splicing alterations linked to splicing factor (SF) activity.
• Elucidating the precise impact of each SF on splicing is crucial for understanding transcriptomic variation.

Purpose of the Study:

• To introduce KATMAP, an interpretable regression model for analyzing transcriptomic splicing changes.
• To model splicing alterations by integrating SF binding data and RNA processing changes.

Main Methods:

• KATMAP utilizes SF perturbation RNA-seq data and SF binding motifs as inputs.
• It employs a regression approach to learn SF position-specific regulatory activity and predict target genes.
• Pretrained models on ENCODE SF knockdown data are available within the KATMAP software.

Main Results:

• KATMAP provides detailed descriptions of SF regulatory activity and identifies predicted SF targets.
• The model can predict SF regulation and cis-elements at individual exons.
• It effectively distinguishes direct SF targets from indirect effects in RNA-seq data.

Conclusions:

• KATMAP offers a powerful tool for interpreting RNA-seq data and understanding SF roles in splicing.
• The model can infer causative SFs from clinical RNA-seq data.
• KATMAP findings can guide the design of splice-switching antisense oligonucleotides.