Customer Reviews

Computational Molecular Biology: An Algorithmic Approach (Computational Molecular Biology)

5 star:		(2)
4 star:		(4)
3 star:		(1)
2 star:		(0)
1 star:		(0)

 

 

The title is somewhat misleading because the book is primarily devoted to combinatorial methods that could be used in genome sequencing and genomics. The selection of methods is arbitrary and does not seem to be dictated by either pedagogical or scientific vision. It mainly reflects the author's own work and interests. Contrary to what the editorial review states I find this text quite abstract and formal. I like the book very much but I don't think it should be recommended to the beginners in computational biology. Readers who have a taste for mathematics and a strong background in combinatorics could benefit the most from reading this book. Anybody who looks for a textbook-level guidance in computational biology should probably rely on better designed texts such as Don Gusfield's "Algorithms on strings trees and sequences" or "Biological sequence analysis" by Durbin and co-authors. However, the readers who are interested in mathematics behind designs of DNA arrays (chapter 5) or in mathematical treatment of genome rearrangements (chapter 10) should certainly read this book in detail.

If you want to understand what is INSIDE those nice software tools available to molecular biologists and now on the web you have to study this book. It's a little more advanced than Gusfield's in some aspects, and more research oriented. Of course it does not cover uniformly all areas of computational biology: if you know Pavel's work, it would be very easy to predict the content of the chapters. For example, more than 50 pages are dedicated to genome rearrangement, but only 10 on multiple sequence alignment. Anyway, this is good, because we can learn about alignment from many other books, in particular the one by Gusfield. I strongly recommend this book to anyone interested in this fascinating field of Science.

Pevzner has written a very useful book on bioinformatics algorithms, and one that seems reasonably up to date. The table of contents follows a classic plan: restriction maps, assembly and sequencing, 2- and N- way string comparisons, and analysis of rearrangements. There's a good but brief section on mass spec analysis - unfortunately, that chapter is called "Proteomics" even though the term covers a lot more than MS. Other sections skim the surface of hidden Markov models and Gibbs sampling for finding patterns ("motifs") in DNA.

A few chapters have unusual strengths. The "Conway Equation" gives more insight in analysis of motif significance than other introductory books do. The section in sequence comparison pays a lot more attention to BLAST-like algorithms than other books do, also - modern material you'd normally see only in the journals. Also, the section on rearrangements gives some ideas about using rearrangement data for phylogenetic analysis. That really gives the material meaning. Rearrangements aren't just string operations, they're features of evolution, and they can be compared to each other. No matter what the discussion, Pevzner keeps maintains a readable and enjoyably informal tone.

The book does have some weaknesses, though. It's a bit advanced for an undergrad intro, but bottoms out before the Baum-Welch algorithm, for example. Discussion of microarrays for sequencing seems dated. Pevnzer describes their use in sequencing, a rarity now, but skips their use in functional gneomics, where they are used most often. Illustration style is erratic and many diagrams are oddly stretched (3.5, 5.7, 8.3, and others, some much worse). Formal analysis of the algorithms is weak, but Pevzner somewhat makes up for that with better statistical analysis than many authors give. Also, even though the book was reprinted in 2001, it still estimates 100K genes in the human genome.

This is a good second book, maybe the one to read after Pevzner's newer "Introduction". It covers most of the basics and gives fairly usable pseudocode. Most of all, it always keeps the biology in mind. That, by itself, makes this book stand out.

//wiredweird

As others have noted, the premise that this book is for beginners from either the computational or the biological field is flawed...unless one's definition of beginner is a lot more advanced than mine.

For example even chapter one throws out terms like "recombination" and electrophoresis. without enough explanation for the biology newbie, IMO. Heck, for someone truly new to biology, a bit of time explaining what a chromosome is is probably time well spent.

And for the person coming from a pure biology background, some of the mathematics will definitely be a problem unless they have a decent understanding of combinatorics and discrete mathematics. And that "computational biology without formulas" blurb on the back cover should be read as "not as many formulas as I could have included if I really wanted", rather than "no formulas at all". There are equations galore in this book, rest assured of that.

That said, if a person *does* have the necessary background to make the material accessbile, then the book is definitely worth the purchase. The book's failure is in defining its target audience, not in the material presented.

Dr. Pevzner writes with a very lucid and conversational style about very complex and seemingly inscrutable topics. As a biologist who works primarily with computational tools in the field of genomics, this resource has helped to provide me with more than a rudimentary understanding of the algorithms and logic lurking in the methods of sequence analysis. Explaining dynamic programming to a biologist with rudimentary programming skills is a daunting task. However, his description of sequence alignment algorithms (including dynamic programming) in chapter 6 is quite readable and the information is very accessible. I highly recommend this book if you want a comprehensive understanding of the computational biologists toolkit.

While this is certainly the do-loop of computational biology the reader would question the assertion that this book provides a common link (no pun) between the biologists need for computational expertise and the programmer's need for biological insight. In either case a solid basis in Discrete Mathematics goes along way here (usually a required course for computer science majors). This reader thinks a similar required course in genetics should be made for engineers to reduce their reductionistic tendencies. However the distinction between these lines grows narrower with each new computer chip. None the less the book is well written, and easy to read (as Discrete Math stuff goes). This book is not for beginners in either Combinatorics or genetics and the last part of the book poses many current questions that as the author says, "are just currently being answered". This book already assumes you know about such things as NIH, PDB, Chime, Isis, NCIB, docking, etc. For those less adapt at programming (myself) the following alternatives are fun, useful and to the point. Both trees and networks can be easily set up in MathCad using their built in resource center add-ins for Combinatorics and Set Theory. They also provide a Traveling Salesman routine in Numerical Recipes that can be applied directly to the problems in Pevzner's book. (Although remembering that most optimization algorithms provide only the most probable 100 out of 2 million it is still fun!). Most of the mappings and node process familiar to Discrete Math can be solved using Mathcad and some sort of adjacency matrix combination. (Including the four-color mapping problem). This provides the basis for most nodal mappings. For the more daring the adjacency matrices can be run through Matlab's GUI's decompositions and analyzed using their optimization toolbox. Currently I'm investigating the Hidden Markovian chains using the Frame advance feature of Mathcad applied to 2D cspline- intercept graphing and updating by frame iteration. This book is for the serious student or solid course material in a related field, and while probably not rated in top ten novels of 2000 certainly rates five mouse clicks from this reader.

An excellent book for studying computational molecular biology from an algorithmic perspective. (But if you never took mathematics seriously, you are forewarned.)