Monthly Archives: August 2019

Aug. 19, 2019. Review article about machine learning for predicting chemotherapy response

Posted on by

New review article in Molecular Genetics and Metabolism about predicting responses to chemotherapy:

Multigene signatures of responses to chemotherapy derived by biochemically inspired machine learning

Published: https://doi.org/10.1016/j.ymgme.2019.08.005

Abstract:   Pharmacogenomic responses to chemotherapy drugs can be modeled by supervised machine learning of expression and copy number of relevant gene combinations. Such biochemical evidence can form the basis of derived gene signatures using cell line data, which can subsequently be examined in patients that have been treated with the same drugs. These gene signatures typically contain elements of multiple biochemical pathways which together comprise multiple origins of drug resistance or sensitivity. The signatures can capture variation in these responses to the same drug among different patients.

July 31, 2019. New preprint on biodosimetry in BioRxiv

Posted on by

“Automated Cytogenetic Biodosimetry at Population-Scale”

PK Rogan, R Lu, E Mucaki, S Ali, B Shirley, Y Li, R Wilkins, F Norton, O Sevriukova, D Pham, E Ainsbury, J Moquat, R Cooke,

T Peerlaproulx, E Waller, JHM Knoll

https://www.biorxiv.org/content/10.1101/718973v1 (doi: https://doi.org/10.1101/718973)

Introduction The dicentric chromosome (DC) assay accurately quantifies exposure to radiation, however manual and semi-automated assignment of DCs has limited its use for a potential large-scale radiation incident. The Automated Dicentric Chromosome Identifier and Dose Estimator Chromosome (ADCI) software automates unattended DC detection and determines radiation exposures, fulfilling IAEA criteria for triage biodosimetry. We present high performance ADCI (ADCI-HT), with the requisite throughput to stratify exposures of populations in large scale radiation events.

Methods ADCI-HT streamlines dose estimation by optimal scheduling of DC detection, given that the numbers of samples and metaphase cell images in each sample vary. A supercomputer analyzes these data in parallel, with each processor handling a single image at a time. Processor resources are managed hierarchically to maximize a constant stream of sample and image analysis. Metaphase data from populations of individuals with clinically relevant radiation exposures after simulated large nuclear incidents were analyzed. Sample counts were derived from US Census data. Analysis times and exposures were quantified for 15 different scenarios.

Results Processing of metaphase images from 1,744 samples (500 images each) used 16,384 CPUs and was completed in 1hr 11min 23sec, with radiation dose of all samples determined in 32 sec with 1,024 CPUs. Processing of 40,000 samples with varying numbers of metaphase cells, 10 different exposures from 5 different biodosimetry labs met IAEA accuracy criteria (dose estimate differences were < 0.5 Gy; median = 0.07) and was completed in ~25 hours. Population-scale metaphase image datasets within radiation contours of nuclear incidents were defined by exposure levels (either >1 Gy or >2 Gy). The time needed to analyze samples of all individuals receiving exposures from a high yield airborne nuclear device ranged from 0.6-7.4 days, depending on the population density.

Conclusion ADCI-HT delivers timely and accurate dose estimates in a simulated population-scale radiation incident.

#nuclear #biodosimetry #radiationprotection #iaea #radiation #population#bioinformatics #highperformancecomputing