Dataset Information

Evaluation of combinatorial optimisation algorithms for c-optimal experimental designs with correlated observations.

ABSTRACT: We show how combinatorial optimisation algorithms can be applied to the problem of identifying c-optimal experimental designs when there may be correlation between and within experimental units and evaluate the performance of relevant algorithms. We assume the data generating process is a generalised linear mixed model and show that the c-optimal design criterion is a monotone supermodular function amenable to a set of simple minimisation algorithms. We evaluate the performance of three relevant algorithms: the local search, the greedy search, and the reverse greedy search. We show that the local and reverse greedy searches provide comparable performance with the worst design outputs having variance <10% greater than the best design, across a range of covariance structures. We show that these algorithms perform as well or better than multiplicative methods that generate weights to place on experimental units. We extend these algorithms to identifying modle-robust c-optimal designs.

Supplementary information

The online version contains supplementary material available at 10.1007/s11222-023-10280-w.

SUBMITTER: Watson SI

PROVIDER: S-EPMC10386961 | biostudies-literature | 2023

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Evaluation of combinatorial optimisation algorithms for c-optimal experimental designs with correlated observations.

Watson Samuel I SI Pan Yi Y

Statistics and computing 20230729 5

We show how combinatorial optimisation algorithms can be applied to the problem of identifying c-optimal experimental designs when there may be correlation between and within experimental units and evaluate the performance of relevant algorithms. We assume the data generating process is a generalised linear mixed model and show that the c-optimal design criterion is a monotone supermodular function amenable to a set of simple minimisation algorithms. We evaluate the performance of three relevant ...[more]

PMID: 37525745

Similar Datasets

Project description:BackgroundThe spread of infectious diseases in populations is controlled by the susceptibility (propensity to acquire infection), infectivity (propensity to transmit infection), and recoverability (propensity to recover/die) of individuals. Estimating genetic risk factors for these three underlying host epidemiological traits can help reduce disease spread through genetic control strategies. Previous studies have identified important 'disease resistance single nucleotide polymorphisms (SNPs)', but how these affect the underlying traits is an unresolved question. Recent advances in computational statistics make it now possible to estimate the effects of SNPs on host traits from epidemic data (e.g. infection and/or recovery times of individuals or diagnostic test results). However, little is known about how to effectively design disease transmission experiments or field studies to maximise the precision with which these effects can be estimated.ResultsIn this paper, we develop and validate analytical expressions for the precision of the estimates of SNP effects on the three above host traits for a disease transmission experiment with one or more non-interacting contact groups. Maximising these expressions leads to three distinct 'experimental' designs, each specifying a different set of ideal SNP genotype compositions across groups: (a) appropriate for a single contact-group, (b) a multi-group design termed "pure", and (c) a multi-group design termed "mixed", where 'pure' and 'mixed' refer to groupings that consist of individuals with uniformly the same or different SNP genotypes, respectively. Precision estimates for susceptibility and recoverability were found to be less sensitive to the experimental design than estimates for infectivity. Whereas the analytical expressions suggest that the multi-group pure and mixed designs estimate SNP effects with similar precision, the mixed design is preferred because it uses information from naturally-occurring rather than artificial infections. The same design principles apply to estimates of the epidemiological impact of other categorical fixed effects, such as breed, line, family, sex, or vaccination status. Estimation of SNP effect precisions from a given experimental setup is implemented in an online software tool SIRE-PC.ConclusionsMethodology was developed to aid the design of disease transmission experiments for estimating the effect of individual SNPs and other categorical variables that underlie host susceptibility, infectivity and recoverability. Designs that maximize the precision of estimates were derived.

Dataset Information

Evaluation of combinatorial optimisation algorithms for c-optimal experimental designs with correlated observations.

Supplementary information

Publications

Evaluation of combinatorial optimisation algorithms for c-optimal experimental designs with correlated observations.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets