Abstract
There is increasing interest in developing diagnostics that discriminate individual mutagenic mechanisms in a range of applications that include identifying population specific mutagenesis and resolving distinct mutation signatures in cancer samples. Analyses for these applications assume that mutagenic mechanisms have a unique relationship with neighboring bases that allows them to be distinguished. Direct support for this assumption is limited to a small number of simple cases, e.g. CpG hypermutability. We have directly evaluated whether the mechanistic origin of a point mutation can be resolved using only sequence context for a more complicated case. We contrasted mutations originating from the multitude of mutagenic processes that normally operate in the mouse germline with those induced by the potent mutagen N-ethyl-N-nitrosourea (ENU). The considerable overlap in the mutation spectra of these two samples make this a challenging problem. Employing a new, robust log-linear modelling method, we demonstrate that neighboring bases contain information regarding point mutation direction that differs between the ENU-induced and spontaneous mutation classes. A logistic regression classifier proved to be substantially more powerful at discriminating between the different mutation classes than alternatives. Concordance between the feature set of the best classifier and information content analyses suggest our results can be generalized to other mutation classification problems. We conclude that machine learning can be used to build a practical classification tool to identify the mutation mechanism for individual genetic variants. Software implementing our approach is freely available under the BSD 3-clause license.
Author Summary Mutations are a fundamental contributor to developing diversity in biological capabilities. There are a multitude of processes that affect how DNA is damaged and whether the damage is repaired in such a way as to produce a mutation. In some instances, mutation is a key adaptive feature of normal biological tissue development. For instance, some immune cells employ enzymatic machinery to elevate the mutation rate to generate additional diversity in the molecules responsible for recognizing pathogens. Mutation is also key to abnormal developmental processes, such as those characterizing cancers. For these and other applications, knowing the mechanisms by which individual mutations arise would enhance our understanding of the biology. Here, we have established and exploited the existence of a relationship between the DNA sequence immediately flanking point mutations and the mechanism of mutation. Using machine learning techniques, we demonstrate how to distinguish mutations that occurred by normal cellular processes from those induced by a potent mutagen. The resulting classifiers showed very strong performance in a manner that was robust. Our results demonstrate the potential for high resolution determination across the genome of where individual mutagenesis mechanisms have operated.