Poster Presentation 43rd Lorne Genome Conference 2022

Convolutional neural networks for genomic and transcriptomic data analysis (#103)

Tansel Ersavas 1 2 , Martin A Smith 3 , John Mattick 1
  1. Biotech & Biomolecular Science, University of New Sout Wales, Kensington, NSW, Australia
  2. Garvan Institute of Medical Research, Darlinghurst, NSW, Australia
  3. Faculté de médecine - Département de biochimie et médecine moléculaire, Universite de Montreal, Montreal, Quebec, Canada

Biological data derived from whole genome sequencing or single cell analysis using short- or long-range sequencers can be challenging to analyse with existing statistical or machine learning tools and methods. 

In this poster we demonstrate how big datasets can be analysed with Convolutional Neural Networks by turning datasets to image like constructs with examples from our previous and current work. We illustrate DNA, RNA, and methylation data can be analysed without much pre-processing or filtering. We also share some results obtained from such analysis.

Deep learning, especially Convolutional Neural Networks are under-utilised tools for biological data analysis, and we believe by pioneering and demonstrating their use in genomic and other biological data will open new doors and lead to new insights in analysis of big datasets