Monday, December 7, 2015
Recent book reviews
In one of the earliest blog posts (Reviews of recent books) I provided links to some book reviews. Recently, a few have appeared for Dan Gusfield's book: ReCombinatorics: the algorithmics of ancestral recombination graphs and explicit phylogenetic networks (2014. The MIT Press, Cambridge, MA).
In addition to the three endorsements that appear as part of the publisher's blurb, a number of independent book reviews have appeared since its publication:
(2014) Computing Reviews Review#143064.
Michael Sanderson (2015) Quarterly Review of Biology 90: 344-345.
Luay Nakhleh (2015) SIAM Reviews 57: 638-642.
From the mathematical point of view, the reviews make it clear that this book is necessary because networks are very much part of the fringe of the computational sciences. Indeed, the challenge is to convince mathematicians that interesting mathematical problems exist with the the study of networks. In this sense, the main limitation of the book is its focus on the parsimony criterion for optimization, rather than statistical approaches to inference, which play such a large part in phylogenetic analyses.
From the biological point of view, the principal issue seems to be the reliance of the book on the infinite sites model, which does not currently have wide applicability in phylogenetics (eg. mostly in population studies such as haplotype inference and association mapping).
The ultimate goal for both computational end biological scientists is working out how to include recombination in the framework of other types of phylogenetic networks. A basic assumption of many phylogenetic analyses is that there has been no recombination. This is because recombination can destroy much of the evidence left by historically preceding processes, so that neither genotype nor phenotype data can reveal patterns and processes that pre-date the recombination events. In this sense, recombination becomes the reticulation process, rather than processes like hybridization or introgression.