Webinar: Run-Length Encoding the Positional Burrows–Wheeler Transform (livestream replay)

Webinar: Run-Length Encoding the Positional Burrows–Wheeler Transform

(29. 5. 2026)

Lecturer: Davide Cozzi, University of Milano-Bicocca

In this webinar, we focus on the Positional Burrows-Wheeler Transform (PBWT), an indexing data structure introduced by Durbin to efficiently query large haplotype panels.

While traditional PBWT methods suffer from high memory consumption, we developed the mu-PBWT, a run-length encoded variant inspired by the r-index that requires sublinear space.

By leveraging succinct data structures and run-length encoding, we show how the mu-PBWT solves matching problems, such as SMEMs and Minimal Positional Substring Cover (MPSC), in O(r) space.

Finally, we highlight with experimental results how this approach enables efficient analysis and scalable querying of population-scale genomic datasets, such as the UK Biobank and the 1000 Genomes Project, even on a commodity laptop.

Livestream replay → https://youtu.be/giWm8ZXutII?si=JXdhc9Fi1v5JY3KW