Poster Australian Marine Sciences Association 2022

Staying afloat in the data ocean: Machine learning in aquaculture transcriptomics (#508)

Megan M Soria 1 , Mark Ziemann 1 , Dean R Jerry 2 , Erica V Todd 1
  1. Deakin University, Waurn Ponds, VIC, Australia
  2. James Cook University, Townsville , QLD, Australia

Transcriptomics is a powerful tool to understand patterns of gene expression, gene function, and the molecular basis of growth, reproduction, nutrition, and disease resistance of marine organisms making it an invaluable tool for aquaculture. The rise of high-throughput sequencing technologies have amassed an ever growing ocean of data as whole transcriptomes for various tissues, species and states accumulate in freely accessible databases. The challenge of analysis is becoming apparent as this data wave requires specialised methods and resources like high performance computing and advanced programming skills. Untapped novel insights remain hidden as the problem of massive data evaluation persists. Machine learning approaches hold great promise for massive biological data analysis, and a few applications are starting to be seen in marine transcriptomics. We reviewed the literature to identify recent advances in transcriptomics analysis using machine learning in aquaculture over the past five years. We identified the commonly used approaches in growth and nutrition, reproduction, disease, and ecology. We investigated which experimental designs are most effective, the effects of sample size on the resulting data and weighed them against economic limits. We singled out the top three approaches and noted their biggest advantages and critiques. Here we present a state of the field and provide an overview of emerging trends. We argue that machine learning can be an invaluable tool in the aquaculture researcher’s toolbelt and can help mine insights from the data ocean.