Bayesian Detection of Abnormal ADS in Mutant Caenorhabditis elegans Embryos

TL;DR

This paper introduces a Bayesian statistical method to detect abnormal cell division timing in mutant Caenorhabditis elegans embryos by comparing it to wild-type, accounting for small sample sizes and correlations with mother cell lifespan.

Contribution

The study develops a semiparametric Bayesian quantile regression approach to identify abnormal division timings, improving detection in small-sample mutant data.

Findings

Method accurately detects abnormal ADS in simulations.

Gene enrichment analysis supports real data results.

Approach accounts for correlation with mother cell lifespan.

Abstract

Cell division timing is critical for cell fate specification and morphogenesis during embryogenesis. How division timings are regulated among cells during development is poorly understood. Here we focus on the comparison of asynchrony of division between sister cells (ADS) between wild-type and mutant individuals of Caenorhabditis elegans. Since the replicate number of mutant individuals of each mutated gene, usually one, is far smaller than that of wild-type, direct comparison of two distributions of ADS between wild-type and mutant type, such as Kolmogorov- Smirnov test, is not feasible. On the other hand, we find that sometimes ADS is correlated with the life span of corresponding mother cell in wild-type. Hence, we apply a semiparametric Bayesian quantile regression method to estimate the 95% confidence interval curve of ADS with respect to life span of mother cell of wild-type individuals. Then, mutant-type ADSs outside the corresponding confidence interval are selected out as abnormal one with a significance level of 0.05. Simulation study demonstrates the accuracy of our method and Gene Enrichment Analysis validates the results of real data sets.