Bioinformatics Seminars

Bioinformatics Seminar

Time:
Venue: Na

17 November 2020

Na

Towards a statistical framework for cell lineage trees

Damien Hicks
Swinburne University of Technology

A cell lineage tree reveals the hierarchy of decision-making processes in a multicellular organism as it develops from a single fertilized egg into different cell types and organs. Measurements of cell lineage trees have a long history ; beginning with live imaging over a century ago and culminating in the explosion of new molecular techniques over the last few years. Analysis of these trees has been largely by visual inspection ; but interpretation becomes difficult in the presence of significant variability. Here I outline a statistical framework that can identify where lasting decisions are made in noisy lineage trees. Founded on the second-order exchangeability properties of a binary tree ; this approach is related to a suite of statistical and signal processing methods ; including classical ANOVA ; graphical models ; and wavelet transforms. I'll describe a new Bayesian version of the technique and its application to both nearly-invariant worm data and highly-variable mammalian T-cell data.;;


Search past seminars