Niko Beerenwinkel | Search bioRxiv

bioRxiv2016

Tree inference for single-cell data

Understanding the mutational heterogeneity within tumours is a keystone for the development of efficient cancer therapies. Here, we present SCITE, a stochastic search algorithm to identify the evolutionary history of a tumour from noisy and incomplete mutation profiles of single cells. SCITE comprises a exible MCMC sampling scheme that allows the user to compute the maximum-likelihood mutation history, to sample from the posterior probability distribution, and to estimate the error rates of the underlying sequencing experiments. Evaluation on real cancer data and on simulation studies shows the scalability of SCITE to present-day single-cell sequencing data and improved reconstruction accuracy compared to existing approaches.

Cancer Biology↗

bioRxiv2015

Latent epistatic interaction model identifies loci associated with human working memory

BackgroundEpistatic interactions among genomic loci are expected to explain a large fraction of the heritability of complex diseases and phenotypic traits of living organisms.\n\nAlthough epistasis detection methods are continually being developed, the current state of the art is exhaustive search methods, which become infeasible when the number of analyzed loci is large.\n\nResultsWe develop a novel latent interaction-based selection method for polymorphic loci as the first stage of a two-stage epistasis detection approach. Given a continuous phenotype and a single-nucleotide polymorphism (SNP), we rank the SNPs according to their interaction potential. When tested on simulated datasets and compared to standard marginal association and exhaustive search methods, our procedure significantly outperforms main-effect heuristics, especially in the presence of linkage disequilibrium (LD), which is explicitly accounted for in our model. Applied to real human genotype data, we prioritized several SNP pairs as candidates for epistatic interactions that influence human working memory performance, some of which are known to be connected to this phenotype.\n\nConclusionsThe proposed method improves two-stage epistasis detection. Its linear runtime and increased statistical power contribute to reducing the computational complexity and to addressing some of the statistical challenges associated with the genome-wide search for epistatic loci.

Bioinformatics↗