Functional linear mixed effects model for mode-of-action clustering in toxicity assessment

• Author / Creator

  Ma, Tiantian

• Real Time Cell Analysis (RTCA) technology is used to monitor cellular changes continuously
  over the entire exposure period to chemicals. In RTCA system, chemicals with dierent concen-
  trations are applied and time-dependent concentration response curves (TCRCs) are generated. In
  this thesis, we aim to study the mode of action (MOA) of tested chemicals by extracting important
  information from TCRCs and then do MOA clustering. In order to reduce the number of param-
  eters to be estimated when tting the data with limited sample size and high dimension, linear
  mixed eects models are applied by considering chemicals as random eects. The estimated and
  predicted coecients from individual curves can be plugged into K-means and Self-organising maps
  to do clustering. Estimating curves using dierent functional bases corresponds to linear transfor-
  mations of the data, and obtains information from various aspects of the curves. In this thesis,
  two dierent functional bases are used when tting linear mixed models. The rst model is based
  on functional principal components, which can stretch the data on a few directions that contain
  almost all the information. The two largest clusters, cluster 1 and cluster 10, can be separated
  with 88:24% accuracy rate on only two primary basis functions. According to the shape of the two
  functions and the coecients distribution on them, we can depict the primary dierence between
  clusters in terms of overall shape and local features. To detect the primary time intervals where the
  dierence lies in, the other basis applied to mixed model is B-spline basis because dierent splines
  are dominant in dierent time intervals. The coecients of spline basis perform well as input in
  both binary and multi-cluster clustering, with the clustering accuracy rate in the range of 81:82%
  to 86:54%. Those clustering results can be obtained by only using 1 to 2 primary directions in
  terms of time interval and concentration level, which is helpful to establish targeted experiments
  for further toxicants study.

• Subjects / Keywords

  • Time-dependent concentration response curves; Mode of action; Linear mixed eect model; Functional principal component analysis; B-spline; K-means; Self-organising map

• Graduation date

  Fall 2020

• Type of Item

  Thesis

• Degree

  Master of Science

• DOI

  https://doi.org/10.7939/r3-8b8p-jv51

• License

  Permission is hereby granted to the University of Alberta Libraries to reproduce single copies of this thesis and to lend or sell such copies for private, scholarly or scientific research purposes only. Where the thesis is converted to, or otherwise made available in digital form, the University of Alberta will advise potential users of the thesis of these terms. The author reserves all other publication and other rights in association with the copyright in the thesis and, except as herein before provided, neither the thesis nor any substantial portion thereof may be printed or otherwise reproduced in any material form whatsoever without the author's prior written permission.