slack---line

Better late than never...

2007-09-18T11:56:00.000Z

Perl scripts & LaTeX

2007-04-18T11:39:00.000Z

Not really related to any news, but I've written a couple of perl scripts for formatting the output from two tagging programs. In both cases the scripts take the standard output and generate L_aT_eX formatted tables for subsequent use.

Tagger (de Bakker et al 2005) is implemented in the program Haploview, and uses pair-wise LD to tag regions. The script for re-formatting the output is written in perl and is available here.

ldSelect (Carlson et al 2004) is an alternative method of tagging SNPs, but also based on LD. The output format from this is (in my opinion) a little awkward as the output uses the base-pair positions (mainly because the input file format is Prettybase). Thus in addition to the output file this script requires the user to provide a tab-delimited text file of the markers and their base-pair positions and will convert the base-pair positions of markers to their rs numbers in the resulting latex table. The script is again written in perl and is available here.

References

Carlson, C. S., Eberle, M. A., Rieder, M. J., Yi, Q., Kruglyak, L., Nickerson, D. A. (2004) Selecting a maximally informative set of single- nucleotide polymorphisms for association analyses using linkage disequilibrium American Journal of Human Genetics 74:106-120 PDF on PubMed

de Bakker, P. I. Yelensky, R. Pe'er, I. Gabriel, S. B. Daly, M. J., Altshuler, D. (2005) Efficiency and power in genetic association studies. Nature Genetics 37:1217-1223

To test or not to test that is the question

2007-01-09T01:42:00.000Z

A recent paper by Zou & Donner (2006) questions whether testing for Hardy-Weinberg equilibrium (HW-eqm)case-control association studies is a viable strategy.

The main point they make is that genotyping error is unlikely to be detected by testing for departure from HW-eqm. This is for a number of reasons, the assumptions underlying HW-eqm, and secondly (perhaps more importantly) that by performing a two stage analysis of screening for deviations for HW-eqm and then only taking forward markers which do not deviate inflates the Type 1 Error rate, because the p-value can not be interpreted as evidence that alleles are independent (i.e. HW-eqm holds).

This is of course appealing as it reduces the computational burden (particuarly when performing whole genome screens by association where multiple testing becomes a big problem), but also because a large proportion of associations are seen where loci do deviate from HW-eqm.

The authors propose a new test for adjusted χ² test based on the difference in variance of the estimated allele frequencies in cases and controls which essentially is essentially the same as a Cochran-Armitage trend test (Sassieni, 1977). The power and performance of this test is discussed in a upcoming paper in Annals of Human Genetics (Ahn et al 2007).

So the upshot of all of this is that as a first pass screen your probably better of using a robust test rather than worrying about deviations from Hardy-Weinberg equilibrium.

References and Links

Zou GY, Donnet A (2006) The Merits of Testing Hardy-Weinberg Equilibrium in the Analysis of Unmatched Case-Control Data: A Cautionary Note Annals of Human Genetics 70:923-933
Ahn K, Haynes C, Kim W, Fleur RS, Gordon D, Finch SJ (2007) The Effects of SNP Genotyping Errors on the Power of the Cochran-Armitage Linear Trend Test for Case/Control Association Studies. Annals of Human Genetics AOP:
Sassieni PD (1997) From genotype to genes: doubling the sample size 53:1253-1261

Fast Genome Wide Association Analysis

2007-01-01T11:29:00.000Z

Recently a new R package that provides an interface to running PBAT in parallel was published in Bioinformatics.The package is described in the Open Access article here.

This is the first package that I've come across that provides an easy interface to analyse the massive numbers of data that are being generated by the new genotyping arrays such as Illumina's bead station and Affymetrix' SNP chip. Basically the data is split up and small chunks analysed on each processor, with a seperate instance of PBAT running on eahc processor. The graphical R front end allows users and easy way to select their analysis option and then handles splitting the data, sending off jobs and amalgamating the results when they are completed.

Refernces and Links

Hoffmann T, Lange C (2006) P²BAT: a massive parallel implementation of PBAT for genome-wide association studies in R. Bioinformatics:22:3103-3105
PBAT
CRAN - Download P²BAT R-package (and many other R packages)
R
http://people.fas.harvard.edu/~tjhoffm/pbatR.html - Detailed instructions on installing and using P²BAT

2006-10-23T01:57:00.000Z

A few special issues of journals have focused on Statistical Genetics recently.

The journal Statistics in Medicine is celebrating its 25th Anniversary and recently published a special issue focusing on statistical genetic analyses. A number of articles are published on topics from genome-wide analyses, estimation of genetic and environmental components in specific diseases, multiple testing and more. The full contents can be read here.

The October issue of Nature Reviews Genetics has a series of interesting articles on statistical genetics. Papers include a reviews of population genetics software, methods of analysing molecular genetic variation, and the analysis of relatedness. There is also a thoughtful perspective paper on the need for longtitudinal analyses for investigating the aetiology of complex diseases. The full contents can be read here.

Genetics Blogs....

2006-09-05T03:30:00.000Z

A Research Highlight in Nature Reviews Genetics higlights the use of the much hyped Web2.0 (wiki's, blogs, social networking sites, interactive localised information etc.) by geneticist's and highlight's a few interesting sites (although manages to miss this one, prehaps that's a bit of motivation to actually get on and write more regularly about interesting developments!!!).

The blogs listed are...

Mendel's Garden
Free Association
Daily Transcript
Genomics Policy
Pahryngula

All a lot more topical than this little blog, but then the authors are probably more motivated to write than I am.

Rub the lamp and make a wish....

2006-04-26T13:27:00.000Z

A recent article by Allen-Brady , Wong & Camp (2006) in BMC Bioinformatics details a new program called PedGenie which provides a java (GUI) interface to performing association and transmission disequilibrium tests (TDT).

Personally I have mixed feelings about GUI interfaces for analysis, firstly they mean that it is hard to make work reproducible, which is an essential part of any investigation (Gentleman 2005). More importantly because people are often very familiar with point and click interfaces they can go ahead and perform their analysis witout necessarily understanding whether it is appropriate to perform their analysis, or even worse they'll perform every possible test they can and think that because they've found something "significant" they have found an important front. This may come acorss as elitist, but its really not meant to be, since when performing statistical analysis one needs to be very careful and precise about what you are doing.

On the flip side GUI's provide a relatively easy interface to what can be a sometimes impenetrable field, since there is often a steep learning curve to becoming proficent in a given statistical package, OS or scripting language to the extent that one can perform their analysis in a reproducable manner.

At the end of the day its about finding a balance between needing to getting things done, and hey, if it means theres one less person banging on my door needing their hand held to perform their analysis then its not all bad!

References

Allen-Brady K, Wong J, Camp NJ (2006) PedGenie: an analysis approach for genetic association testing in extended pedigrees and genealogies of arbitrary size BMC Bioinformatics 7:209
Gentlemean R (2005) Reproducible Research: A Bioinformatics Case StudyStatistical Applications in Genetics and Molecular Biology 4:2
:

(m)utterings....

2006-04-12T13:48:00.000Z

A rather provocative article appeared in the European Journal of Human Genetics recently from Terwilliger & Tero Hiekkalinna (2006) "An utter refutation of the 'Fundamental Theorem of the HapMap'". The article provides a mathematical proof that the theorem behind the HapMap is flawed.

It is often assumed that the power to detect association is directly proportional to the correlation coefficent (r²) between the disease locus B and the marker locus A, but this is not the case. This makes intuitive sense, for the power of testing a proxy marker to be equal to that of testing the disease polymorphism itself, the allele frequencies must be identical. In such cases p_a = p_b (i.e. r² = 1), if the allele frequencies are not the same the there will be instances when the marker allele (a) occurs with the 'normal' allele B. The paper claims that because this assumption is flawed then it will not be possible to utilise LD mapping to sucessfully map disease polymorphisms. This may be true, but there is evidence that the structure of LD in the human genome is not uniform, there are some regions of strong LD where D' and r² are high, and others where it is very weak. As a consequence there will be some regions where LD mapping will work, and others where it will not.

The wording is very strong in the paper, and is rather acerbic in its tone as exemplified in the title "Utter" refutation and and the inverted commas around the Fundamental Theorm (which to the best of my memory I have never heard of it being referred to as such, although it is implicit in the strategy of the HapMap).

To quote someone else "The devil is in the detail" and what is written above is far from all that is covered in the paper, but as it is freely available at the EJHG web-site, head on over, read it and make up your own mind.

Terwilliger J, Hiekkalinna (2006) An utter refutation of the 'Fundamental Theorem of the HapMap' Eurpoean Journal of Human Genetics 14:426-437.

Usual Suspects?

2006-03-13T03:42:00.000Z

A recent article in Bioinformatics describes a web-based server for prioritising candidate genes within linkage regions. It combines annotations and sequence baed approaches, and would be most useful for people who have large regions of chromosomes to follow up on the basis of linkage studies.

References

Adie EA, Adams RR, Evans KL, Porteus DJ, Pickard BS (2006) SUSPECTS: enabling fast and effective prioritization of positional candidates. Bioinformatics 22:773-774

Don't over estimate the power....

2006-03-06T23:05:00.000Z

A lucid example of why one should not treat estimated haplotypes as observed haplotypes has been provided by David Curtis and Pak Sham in a letter to the American Journal of Human Genetics.

They demonstrate through a simple example, and simulations that the method of Zhao et al's (2005) entropy based approach, and that of Sham and Curtis (1995) both of which essentially treat estimated haplotypes as observed, are anticonservative (i.e. over estimate any association/effect).

However in the same issue Zhao et al (2006) provide a counter example , and claim that the range of haplotype frequencies that have been investigated to date is too small, and suggest that further theoretical work is required on this area.

References

Curtis D, Sham P (2006) Estimated Haplotype Counts from Case-Control Samples Cannot Be Treated as Observed Counts. American Journal of Human Genetics 78:729-730

Zhao J, Boerwinkle E, Xiong M (2006) Impact of Haplotype-Frequency Estimation Error on Test Statistics in Association Studies. American Journal of Human Genetics 78:728-729

Zhao J, Boerwinkle E, Xiong M (2005) An entropy-based statistic for genomewide association studies American Journal of Human Genetics 77:27-40

What have we 'Ear'?

2006-02-21T01:32:00.000Z

An AOP publication in Nature Genetics claims to have identified a gene involved in the secretion of wet earwax (Yoshiura et al). Quite why they chose to investigate this trait is beyond my understanding (after all I feel that auto-immune and cardio-vascular diseases and cancers have a far larger impact on human health world-wide than the need for the occasional ear syringe), particularly given that it is mendelian and not polygenic in nature, despite the contentious nature of the association with breast-cancer (which is likely to only be pure co-incedence).

The claim that this is "...the first example of DNA polymorphism determining a visible genetic trait" is I feel somewhat grandiose and misled as there are innumerable polmorphims which have a direct effect on the phenotype of individuals who carry them, for example the ΔF508 mutation in Cystic Fibrosis, or the Huntington's Disease gene.

More interstingly the current issue of Nature Genetics has an article on a sex-xpecific QTL (Quantitative-Trait Locus). In this article they review 17 quantitative traits (including systolic and diastolic blood pressure, triglycerides, serontin levels, LDL and HDL levels) in the Hutterite's to see if there are any sex-specific effects. They found that 11 of these showed differences between the sexes at a significance level of p < 0.05 (although really they should have applied a Bonferroni correction and set the significance level to 0.0029 at a bare minimum assuming all the traits were independent). Heritability differed between sexes in some cases as well.

This obviously has important implications when analysing genetic data for QTL's and sex should therefore always be included in the early stages of the analysis to determine whether it has a significant effect.

Links

Complex Trait Mapping in the Hutterites

DNA databases...

2006-02-07T07:28:00.000Z

Back in April 2005 (21st to be precise), I attended a public debate entitled DNA Profiling and You: Do genetic databases threaten or benefit society? at the Nowgen Centre in Manchester.

I heard that when the databases were initiated in 1999 it was only admisable to take DNA samples from individuals who had been convicted of a crime, but was astounded that since then the laws have been relaxed to such an extent that samples are routinely taken by people who are arrested even if they are acquited of the crime for which they were arrested. This was verified by a friend of mine who was arrested but never charged, yet was obliged (under the threat of force) to provide a DNA sample.

Recent articles in the news are highlighted in this months Nature Reviews Genetics, and highlights not only the racial bias that exists within the UK database, but the rapid expansion that it has undergone (see the article here if you can access it).

I was astounded to read that only ten microsatellites are used for typing individuals as this seems exceptionally small (in the future I may do some simulations to investigate the statistical probabilites of finding matches by chance alone with such a small sample size if time permits).

I do not wish to comment on the moral or ethical implications of collecting such databases, but find it amazing that such databases exist and continue to expand relatively unchecked. It is heinous that samples can be taken and retained freely, particularly in light of the extensive legal and ethical standards that organised medical research projects have to adhere to when applying for and executing medical research.

Links

DNA database continues to swelll (BBC News, 04/01/2006)

Statalist Blog...

2006-02-06T01:48:00.000Z

I use the statistical package Stata and today I came across the blogger feed of the mailing list and thought I'd menion it here.

To view posts to this mailing list simply go here.

Festa(r) & Festa(r) tagging SNPs

2006-02-02T06:06:00.000Z

Yet another SNP tagging algorithm and program has been released, this time from the University of Michigan Center of Statistical Genetics.

Qin et al (2006) Bioinformatics 22:220-225 propose tagging on the basis of the pairwise LD measure r² and have developed a fast and efficent algortihm which breaks genomic regions down into sections, allowing for faster computation of tagging SNPs has been developed to carry this out.

It avoids the potential drawback of multimarker approaches which may not perform well when genotyping fails.

This is a nice simple approach to increasing efficency, since there is little point in attempting to tag SNPs over long distances.

Jump in the pool...

2006-01-17T08:12:00.000Z

An interesting article in Nature Genetics by Skol et al (v38:209-213)investigates different strategies when performing Whole Genome Screens by Association (WGSA) using a two stage design whereby a subset of the total cohort of cases and controls are typed for all markers across the genome (perhaps with Affymetrix's 100k or 500k SNP chip or Perlegen's bead arrays), and a subset of associated markers are selected for follow up in the remaining samples.

The paper shows that irrespective of the proprtion split made in the cohort, or the number of markers that are to be followed up, the most powerful strategy for detecting effects is to analyse the first (screening) and second (replication) cohorts jointly instead of trying to replicate associations in the second cohort alone.

Intuitivevly this makes sense since you are including individuals who have already demonstrated association from the first stage, and seeing if the inclusion of the second cohort reinforces the effect or dissimulates it.

Personally quite interesting since I'm currently involved in such a study.

Reference

Skol AD, Scott LJ, Abecasis GR, Boehnke M (2005) Joint analysis is more efficent than replication-based analysis for two-stage genome-wide association studies. Nature Genetics AOP

Donate your idle CPU time to medicine...

2005-12-30T13:12:00.000Z

Two worthwhile distributed computing projects that I came across today are Folding@home and Genome@home.

The former seeks to understand the way in which proteins secondary and tertiary structures are formed, whilst the later seeks to use this information to derive novel genes (and hence proteins) for application in theraputic medicines.

If you have a computer on your desk, even if you use it throughout the day there is likely to be time when its full CPU capacity is not being used, and by downloading on of the clients you can help solve some of the computational problems that exist within medical research.

Lost hunting for genes?

2005-12-29T23:43:00.000Z

The current issue of Nature Genetics has an interesting article entitled "A road map for efficient and reliable human genome epidemiology", which champions the notion that researchers need to co-ordinate their work.

One key point the authors make is that there is still a publication bias which favours positive associations. This clear bias detracts from the scientific quality of the area as a whole, and runs the risk of engendering data dredging.

A number of online databases are also listed (many of these and more can be found here in the Links section of my web-site)...

Human Genetic Epidemiology Network (HuGENet)
(UK HuGENet Coordinating Centre)
Genetic Association Database
PharmGKB: The Pharmacogenetics and Pharmacogenomics Knowledge Base
STROBE statement

Common disease variants are invariant

2005-12-20T11:55:00.000Z

A recent article in The American Journal of Human Genetics by Lohmueller et al shows that SNP variants involved in 'common' diseases tend not to vary considerably between the two populations they compared.

This is good news in the world of gene hunting since it suggests that these variants arose prior to the divergence of human populations, and that they variants are therefore common to most human populations. This is not to say that genetic heterogeneity does not exist, since there are likely to be other mutations that influence disease that are population specific, but it is encouraging since it means that it should be reasonably easy to replicate associations with some of the common variants (indeed that was one of the criteria for selection used in the paper).

Proceedings of Genetic Analysis Workshop 14

2005-12-14T10:45:00.000Z

The summary papers for the 14th Genetic Analysis Workshop have been published as Supplementary issue to the journal Genetic Epidemiology.

These papers are summaries for each of the working groups that were formed as part of the workshop. Individual work from each members of the group are due to be published as a supplement in BMC Genetics in the not too distant future.

(I attended this meeting and was part of group 7).

Interesting Blogs

2005-12-05T18:57:00.000Z

Nothing much to add today, but here are a few links to other blogs readers may find interesting.

Free Association is the Nature Genetics blog.

And after reading the above blog Paul Myer's blog Pharyngula would be of general interest as well.

When I've got more time I shall add a blog links section to the right.

R for Statistical Genetics

2005-12-04T08:57:00.000Z

R is becoming increasingly popular as a tool for statistical genetics. There are already a number of packages available, and these are detailed at the CRAN Task View: Statistical Genetics.

R has also been extended considerably as part of the Bioconductor project which develops tools for bioinformatics and micro-array analyses.

Pan autoimmune genes...

2005-12-03T08:25:00.000Z

Okay so I guess I should start adding stuff to this blog.

Theres an interesting article on the role of PTPN22, which shows numerous associations with autoimmune diseases. The article is in Nature Genetics and can be read
here.

Welcome

2005-12-02T15:03:00.000Z

Welcome to my blog!

I work in the fast-moving (exciting?) world of human genetics and am a genetic statistician.

I intend to use this page to detail useful information about statistical genetics, so if you work in the field as well you may find it interesting, if you don't then I hope you enjoy reading it anyway.

slack---line

Better late than never...

Perl scripts & LaTeX

References

To test or not to test that is the question

References and Links

Fast Genome Wide Association Analysis

Refernces and Links

Genetics Blogs....

Rub the lamp and make a wish....

References

(m)utterings....

Usual Suspects?

References

Don't over estimate the power....

References

What have we 'Ear'?

Links

DNA databases...

Links

Statalist Blog...

Festa(r) & Festa(r) tagging SNPs

Jump in the pool...

Donate your idle CPU time to medicine...

Lost hunting for genes?

Common disease variants are invariant

Proceedings of Genetic Analysis Workshop 14

Recommended Reading

Interesting Blogs

R for Statistical Genetics

Pan autoimmune genes...

Welcome