Genome-wide Association Analysis of Blood Smear Imaging Phenotypes

Peripheral blood smears from 30,000 healthy blood donors participating in the COMPARE study have been digitally imaged with the Sysmex CellaVision DM96 system. We develop automatic algorithms to derive cellular phenotypes from these smear images, which are to be used in follow-on genetic and aetiological analyses. A fast and accurate algorithm to segment the white blood cells and platelets from sub images containing single cells per image has been prototyped. A classification algorithm to identify the subtypes of normal white blood cells, artifacts and the white cells with sample age related morphological changes has also been developed and applied to the COMPARE study images. Morphological features including measures of cell granularity, size, shape and colour intensity are extracted from the segmented and classified images. We focus on the statistical distribution of the morphological features instead of the univariate cell trait considered in ordinary genome-wide association analysis (GWAS), and propose a new method for the GWAS of distribution phenotypes. A joint modelling framework that combines functional principal component analysis and linear mixed effects models is proposed for the new GWAS and we hope that it will give some insights into the relationship between the genetic components and the distribution of the white cell morphological features.

Join Zoom Meeting
https://maths-cam-ac-uk.zoom.us/j/96932950870?pwd=b1o2c2UxckVITlRlazJzY0laRmVHZz09
Meeting ID: 969 3295 0870
Passcode: DRHjehPj

• Speaker: Chuoxin Ma
• Wednesday 27 October 2021, 13:00–14:00
• Venue: Virtual (see abstract for Zoom link).
• Series: CMIH Hub seminar series; organiser: J.W.Stevens.