Evaluating the Accuracy of Consensus Nanopore Squiggles Generated by Dynamic Time Warp Barycentre Averaging (DBA)

Michael Smith1, Rachel S. L. Chan2, Paul Gordon3

  • 1University of Calgary
  • 2Electrical and Computer Engineering, University of Calgary
  • 3Bioinformatics, Alberta Childrens’ Hospital Research Institute,

Details

08:30 - 08:45 | Wed 24 Jul | M3 - Level 3 | WeA15.1

Session: Bioinformatics - Bioinformatics for Health Monitoring

Abstract

Picoamperage signals are generated as each nucleotide of a DNA or RNA molecule is ratcheted through a nanosequencer’s nanopores by motor proteins. These are segmented into squiggle representations of the molecule. It has been suggested that applying dynamic time warp Barycentre Averaging (DBA) to multiple noisy squiggles can generate a lower noise, less-distorted, consensus signal that retains the key squiggle characteristics that would be distorted by other averaging approaches. We discuss experimental results obtained when developing DBA consensus signals from squiggles produced by an Oxford MinION nano-sequencer squiggle convertor during an Enolase study.Metrics are proposed to identify differences between the known gold standard and consensus signals, and the level of self-consistency between consensus signals developed from noisy squiggles with different length distortions. A number of location-specific differences between the gold and consensus squiggles were identified.