Search found 151 matches

by Nicola Pirastu
Fri Mar 20, 2015 2:30 pm
Forum: GenABEL
Topic: Covariates effects and standard errors?
Replies: 1
Views: 1386

Re: Covariates effects and standard errors?

Hi,

do you mean from the summary statistics of a GWAS or in general?

Best.

Nicola
by Nicola Pirastu
Fri Mar 20, 2015 2:26 pm
Forum: GenABEL
Topic: proportion of heritability explained by a SNP
Replies: 10
Views: 12896

Re: proportion of heritability explained by a SNP

Hi,

the correct formula is actually: r2=chi^2/(N-2+chi^2)

You can find an explaination here:
http://forum.genabel.org/viewtopic.php?f=6&t=760&p=1513&hilit=chi2%2FN#p1513


However I doubt there is a citation for it.

Best.

Nicola
by Nicola Pirastu
Thu Apr 24, 2014 12:32 pm
Forum: GenABEL
Topic: Sex check
Replies: 1
Views: 1308

Re: Sex check

Sure you can,

just put non stringent parameters instead of the defaults. For example you can put maf=0 IBS threshold to 1, callrate to 0 and so on.

This way the other checks will not really be skipped but all ids and SNPs will pass them.

Hope this helps.

Best.

Nicola
by Nicola Pirastu
Thu Apr 24, 2014 12:30 pm
Forum: GenABEL
Topic: [SOLVED] polygenic
Replies: 2
Views: 1678

Re: polygenic

Hi,

another way to read the instructions is:

your_trait~ covariate1+ covariate2....

All covariates are fixed effects.

Does this make it clearer?

Best.

Nicola
by Nicola Pirastu
Sun Feb 09, 2014 8:43 am
Forum: GenABEL
Topic: Testing non-genetic predictors in polygenic mixed model
Replies: 2
Views: 2596

Re: Testing non-genetic predictors in polygenic mixed model

HI, you have basically three choiches: 1) use polygenic in combiantion with the likelihood ratio test, but you will not get standard errors and if you add more than one variable at a time you will not get individual variables p-values. In this case you should first fit a null model like: nullmod=pol...
by Nicola Pirastu
Sun Feb 09, 2014 8:29 am
Forum: GenABEL
Topic: idsubset for qtscore
Replies: 1
Views: 1266

Re: idsubset for qtscore

If I understand correctly what you are trying to do, I think you should try something like:

Code: Select all

vec<-phdata(df)$id[which(phdata(df)$X1 == 1)]


There might be some NAs in $X1 which idsubset seems to not like very much.

Best.

Nicola
by Nicola Pirastu
Sun Feb 09, 2014 8:26 am
Forum: GenABEL
Topic: [Solved] lambda estimate is unreliable
Replies: 4
Views: 1950

Re: Number of observation < 100, lambda estimate is unreliab

Hi,

the number of observation regarding lambda is referred to the number of SNPs not the number of person.
Lambda is calculated only on pvalues and the number of people is irrelevant.
So it's just telling you that lambda is unreliable since you have less than 100 SNPs.

Best.

Nicola
by Nicola Pirastu
Wed Jan 08, 2014 9:27 am
Forum: GenABEL
Topic: [SOLVED] r2fast error
Replies: 4
Views: 2364

Re: r2fast error

Dear Karl, I think you are asking too much from your computer, I think that the problem is that if you try to compute all pairwise comparison it will result in a huge matrix 600.000*600.000 which is really too much. It also does not really make sense to estimate r2 on SNPs either on different chromo...
by Nicola Pirastu
Mon Oct 21, 2013 9:46 am
Forum: GenABEL
Topic: Get eigenvalues for MDS plot
Replies: 1
Views: 1862

Re: Get eigenvalues for MDS plot

Hi, it would be helful to know what you need eigenvalues for. Remember that MDS is not PCA, it is extremelly similar but it is in principal a different thing. So basically if you are trying to see the structure of your population you are doing it correctly but you will not get how much of the total ...
by Nicola Pirastu
Thu Aug 22, 2013 1:17 pm
Forum: ProbABEL
Topic: [SOLVED] Memory mapping error
Replies: 3
Views: 1941

Re: Memory mapping error

Just to add information, without mmscore it works just fine.

N.

Go to advanced search