Toby Johnson | Searcharxiv

arXiv2007

Reciprocal best hits are not a logically sufficient condition for orthology

It is common to use reciprocal best hits, also known as a boomerang criterion, for determining orthology between sequences. The best hits may be found by blast, or by other more recently developed algorithms. Previous work seems to have assumed that reciprocal best hits is a sufficient but not necessary condition for orthology. In this article, I explain why reciprocal best hits cannot logically be a sufficient condition for orthology. If reciprocal best hits is neither sufficient nor necessary for orthology, it would seem worthwhile to examine further the logical foundations of some unsupervised algorithms that are used to identify orthologs.

q-bio.GN↗

arXiv2005

Bayesian Method for Disease QTL Detection and Mapping, using a Case and Control Design and DNA Pooling

This paper describes a Bayesian statistical method for determining the genetic basis of a complex genetic trait. The method uses a sample of unrelated individuals classified into two groups, for example cases and controls. Each group is assumed to have been genotyped at a battery of marker loci using a laboratory effort efficient technique called DNA pooling. The aim is to detect and map a quantitative trait locus (QTL) that is not one of the typed markers. The method works by conducting an exact Bayesian analysis under a number of simplifying population genetic assumptions that are somewhat unrealistic. Despite this, the method is shown to perform acceptably on datasets simulated under a more realistic model, and furthermore is shown to outperform classical single point methods.

q-bio.GN↗