1
1888 BOOK REVIEWS 2. STATISTICS IN GENETICS. M. Elizabeth Halloran and Seymour Geisser (eds), Springer, Surrey, 1999. No. of pages: 260. Price: $ 53.50. ISBN 0-387-98828-9 This volume contains a selection of referred pa- pers from a meeting organized by the Institute of Mathematics and its Applications (IMA) in sum- mer 1997. According to the editors’ preface, the purpose of the meeting was to ‘bring together re- searchers who would not necessarily have attended the same meeting to promote productive and novel discussion’. As a result the subject matter of the papers is rather broad. They are grouped into three sections under the headings of forensics, popula- tion genetics and human genetics, but even within these sections a great variety of topics are covered. The forensics section is perhaps the most ho- mogenous of the three. All three papers in this section are reasonably non-technical, and acces- sible to those with little background. Two articles deal with the use of DNA in criminal trials, and describe some of the many severe problems with the recommendations made by the most recent report of the National Research Council. The problems are clearly explained, and solutions to some are suggested. A third article deals with the aects of population structure on paternity testing. The population genetics section includes only one paper that is strictly on population genetics: a wide-ranging and reasonably non-technical review of coalescent theory and its applications. The other articles are an introductory review of phylogenetic analysis methods (the study of the relationships be- tween species), and a more technical article which takes a phylogenetic approach to detecting linked mutations in HIV. Although it is understandable that these three articles were grouped together, it is perhaps a little unfortunate given that phyloge- netic methods are often applied in contexts where population genetic models might be more appro- priate (the analysis of HIV data being a case in point). The papers on human genetics are certainly the most variable in terms of technicality. There is a review of statistical methods in human genetics, which provides an introduction to the problems of reconstructing the human genome sequence, searching for homology using the BLAST algo- rithm, and nding disease genes through linkage analysis. The review seems to be aimed at math- ematicians who might be interested in getting involved in genetics applications, and gives some advice on which sorts of problems are likely to be of interest in years to come. A rather more specialist review describes recent biological and statistical studies of diseases involving mutations in mitochondrial DNA, and aims to stimulate sta- tistical research in this eld. At the other end of the technical spectrum are papers proving a tri- angle inequality for sib-pair identical by descent probabilities, and examing the eects of popula- tion stratication on the aected-pedigree-member method. These seem likely to be of most interest to specialists in the eld, and I feel unqualied to judge their value. Finally there is a more generally accessible article on analysis of cancer genome variation. As a result of the enormous breadth of material covered, it is dicult to imagine any individual to whom every article in this volume will appeal. On the other hand, there is probably something for ev- eryone who has an interest in statistical genetics, and although the quality of the writing and type- setting is a little uneven, there is certainly enough of interest to justify a place on the library shelf. MATTHEW STEPHENS Department of Statistics 1 South Park Road Oxford OX1 3TG U.K. 3. MULTIPLE REGRESSION: A PRIMER. Paul D. Alli- son, Sage, London, 1999. No. of pages: 220. Price: $ 11.99. ISBN 0-7619-8533-6 Paul Allison has produced an appealing introduc- tory book on multiple regression analysis aimed primarily at students in the social sciences, but also well suited for medical professionals with a basic knowledge of statistics. The overall impression is a very friendly book in a compendium-like layout, which guides the reader carefully into the eld. This has primarily been achieved by a very innova- tive organization of the book with the most essen- tial questions addressed in the early chapters and the less important (and more mathematical) issues relegated to the later chapters. Thus, as opposed to conventional textbooks with their elaborate math- ematical introductions, no readers are likely to be scared o in the beginning. Each chapter is or- ganized as a question and answer session, as in Copyright ? 2001 John Wiley & Sons, Ltd. Statist. Med. 2001; 20:1887–1889

Statistics in genetics. M. Elizabeth Halloran and Seymour Geisser (eds), Springer, Surrey, 1999. No. of pages: 260. Price: £ 53.50. ISBN 0-387-98828-9

Embed Size (px)

Citation preview

Page 1: Statistics in genetics. M. Elizabeth Halloran and Seymour Geisser (eds), Springer, Surrey, 1999. No. of pages: 260. Price: £ 53.50. ISBN 0-387-98828-9

1888 BOOK REVIEWS

2. STATISTICS IN GENETICS. M. Elizabeth Halloranand Seymour Geisser (eds), Springer, Surrey,1999. No. of pages: 260. Price: $ 53.50. ISBN0-387-98828-9

This volume contains a selection of referred pa-pers from a meeting organized by the Institute ofMathematics and its Applications (IMA) in sum-mer 1997. According to the editors’ preface, thepurpose of the meeting was to ‘bring together re-searchers who would not necessarily have attendedthe same meeting to promote productive and noveldiscussion’. As a result the subject matter of thepapers is rather broad. They are grouped into threesections under the headings of forensics, popula-tion genetics and human genetics, but even withinthese sections a great variety of topics are covered.

The forensics section is perhaps the most ho-mogenous of the three. All three papers in thissection are reasonably non-technical, and acces-sible to those with little background. Two articlesdeal with the use of DNA in criminal trials, anddescribe some of the many severe problems withthe recommendations made by the most recentreport of the National Research Council. Theproblems are clearly explained, and solutions tosome are suggested. A third article deals withthe aHects of population structure on paternitytesting.

The population genetics section includes onlyone paper that is strictly on population genetics: awide-ranging and reasonably non-technical reviewof coalescent theory and its applications. The otherarticles are an introductory review of phylogeneticanalysis methods (the study of the relationships be-tween species), and a more technical article whichtakes a phylogenetic approach to detecting linkedmutations in HIV. Although it is understandablethat these three articles were grouped together, itis perhaps a little unfortunate given that phyloge-netic methods are often applied in contexts where

population genetic models might be more appro-priate (the analysis of HIV data being a case inpoint).

The papers on human genetics are certainly themost variable in terms of technicality. There is areview of statistical methods in human genetics,which provides an introduction to the problemsof reconstructing the human genome sequence,searching for homology using the BLAST algo-rithm, and Dnding disease genes through linkageanalysis. The review seems to be aimed at math-ematicians who might be interested in gettinginvolved in genetics applications, and gives someadvice on which sorts of problems are likely tobe of interest in years to come. A rather morespecialist review describes recent biological andstatistical studies of diseases involving mutationsin mitochondrial DNA, and aims to stimulate sta-tistical research in this Deld. At the other end ofthe technical spectrum are papers proving a tri-angle inequality for sib-pair identical by descentprobabilities, and examing the eHects of popula-tion stratiDcation on the aHected-pedigree-membermethod. These seem likely to be of most interestto specialists in the Deld, and I feel unqualiDed tojudge their value. Finally there is a more generallyaccessible article on analysis of cancer genomevariation.

As a result of the enormous breadth of materialcovered, it is diIcult to imagine any individual towhom every article in this volume will appeal. Onthe other hand, there is probably something for ev-eryone who has an interest in statistical genetics,and although the quality of the writing and type-setting is a little uneven, there is certainly enoughof interest to justify a place on the library shelf.

MATTHEW STEPHENSDepartment of Statistics

1 South Park RoadOxford OX1 3TG

U.K.

3. MULTIPLE REGRESSION: A PRIMER. Paul D. Alli-son, Sage, London, 1999. No. of pages: 220. Price:$ 11.99. ISBN 0-7619-8533-6

Paul Allison has produced an appealing introduc-tory book on multiple regression analysis aimedprimarily at students in the social sciences, but alsowell suited for medical professionals with a basicknowledge of statistics. The overall impression isa very friendly book in a compendium-like layout,

which guides the reader carefully into the Deld.This has primarily been achieved by a very innova-tive organization of the book with the most essen-tial questions addressed in the early chapters andthe less important (and more mathematical) issuesrelegated to the later chapters. Thus, as opposed toconventional textbooks with their elaborate math-ematical introductions, no readers are likely to bescared oH in the beginning. Each chapter is or-ganized as a question and answer session, as in

Copyright ? 2001 John Wiley & Sons, Ltd. Statist. Med. 2001; 20:1887–1889