Bayesian Analysis of Frequency of Allelic Loss Data

原文摘要

One objective of allelic-loss studies is to identify chromosomal locations that may harbor tumor-suppressor genes. An instability-selection model has been developed for allelic-loss data in which the loss events are available for each tumor and each marker (allelotypes). In performing pooled analyses of published allelic-loss experiments, however, only summaries of the frequency of allelic loss (FAL) may be available. The instability-selection model can be applied to these summary data, but naive computational approaches are prohibitive. A hidden Markov model (HMM) maximum likelihood approach has recently been proposed for FAL data, but the computation remains challenging. Moreover, precise methods for hypothesis testing and location inference are not available. We propose an alternative Bayesian treatment of the instability-selection model for FAL data. Advantages of the Bayesian approach include the availability of (1) natural imputation approaches to handle missing data, (2) hypothesis testing using Bayes factors, and (3) interpretable posterior intervals for tumor-suppressor locations. We apply our Bayesian approach to four previously reported allelic-loss studies.

摘要来源：OpenAlex 索引摘要（稿件版本未注明）