Pan-transcriptome analysis with a novel computational framework

Transcriptome analysis is used in a vast range of biological experiments to study cell differentiation, medical treatment response, etc. Recently, long-read sequencing technologies have made groundbreaking advances to transcriptome analysis.

They have contributed to identifying transcripts from biological mechanisms believed to cause, e.g., Alzheimer´s and Parkinson´s disease. However, computational methods are required to identify transcripts from the reads. Most methods identify transcripts by the guidance of a reference genome.

The reference genome does not contain biological variation between individuals and causes reference-based methods to miss transcripts with variations absent in the reference.

In genomic analysis, methods capable of representing the biological variation have shown substantial improvement over methods using a reference genome.

This direction has so far not been well-studied in transcriptomic analysis.This project proposes a computational framework that reconstructs and identifies transcripts from long reads without the guidance of a reference genome.

Reconstruction and identification of transcripts are fundamental computational steps for discovering new biological variations and mechanisms that cause disease.

We plan to achieve a reference-free analysis framework by combining our newly developed data structure (strobemers) with a novel representation of transcript variation.