Fifteen Eighty Four

Academic perspectives from Cambridge University Press


Computational phylogenetics for algorithms designers

Tandy Warnow

Phylogenetic trees are used in many biological analyses, including protein structure and function prediction, microbiome analysis, and the inference of human migrations.  Over the last  50  years,  many  statisticians  and  probabilists have made great breakthroughs in both models of sequence evolution and analytical  methods  for  estimating  phylogenies  under these  models,  and  so  have transformed the field of computational methods for phylogeny estimation.  Indeed, the availability of sophisticated computational methods, fast computers and high performance computing (HPC) platforms, and large sequence datasets enabled through DNA sequencing technologies, has led to the expectation that highly accurate large-scale phylogeny estimation, potentially answering open questions about how life evolved on earth, should be achievable.

Yet,  large-scale  phylogeny  estimation  turns  out  to  be  much  more  difficult than expected.  First, all the best methods are computationally intensive, and standard techniques do not scale well to large datasets; massive parallelism helps but does not really address the basic challenge inherent in searching an exponential search space.  Another issue is that the statistical  models  of  sequence  evolution  that  properly  address  genomic  data  are substantially  more complex  than  the  ones  that  model  individual  loci,  and  methods  to estimate genome-scale phylogenies are (relatively speaking) in their infancy compared to methods for single gene phylogenies.  Finally, there is a substantial gap between performance as suggested by mathematical theory (which is used to establish guarantees about methods under statistical models of evolution) and how well the methods actually perform on data – even on data generated under the same statistical models!  Indeed, this gap is one of the most interesting things about doing research in computational phylogenetics, because it means that the most impactful research in the area must draw on mathematical theory (especially probability theory and graph theory) as well as on observations from data.

Computer scientists have brought innovative algorithm design techniques into computational phylogenetics that are dramatically improving the accuracy and scalability of phylogeny estimation.  Many of these new methods are now being used by evolutionary biologists to compute multiple sequence alignments, construct species trees and phylogenetic networks from genome-scale datasets, and make biological discoveries.  It is clear that computer science techniques can- and will- enable breakthroughs in biological discovery for the genome-scale datasets that are being assembled around the world.

Computational Phylogenetics: An Introduction to Designing Methods for Phylogeny Estimation is designed to train the next generation of algorithm developers so that they can develop these new methods and enable these breakthroughs. The book is self-contained, and no biology background is needed. Although the focus is on communicating mathematical foundations and innovative algorithm design, much of the material is accessible to biologists and others who are interested in critically evaluating the scientific literature about phylogeny estimation methods in this post-genome era.

Find out more about Computational Phylogenetics and Tandy Warnow

About The Author

Tandy Warnow

Tandy Warnow is a Founder Professor of Engineering at the University of Illinois, Urbana-Champaign. Her awards include the National Science Foundation Young Investigator Award (199...

View profile >

Latest Comments

Have your say!