|
Post by SLacroix on Feb 5, 2015 14:09:20 GMT
Dear all,
I am working with dog genotype data (26 chromosomes genotyped) and would like to use MLMA-loco. However, GCTA seem to stop the analysis at chromosome 20 (default...human) and doesn't go further when I specify: --autosome-num 26
Did someone run into similar problem? Is there a way to do MLMA-loco with more than 20 chromosomes?
Thanks for your help,
|
|
|
Post by Zhihong Zhu on Feb 8, 2015 11:08:49 GMT
Hi,
I haven't got this problem before. But would you please check the error message if you run the 21st chromosome, specifying "--chr 21"?
Cheers, Zhihong
|
|
|
Post by SLacroix on Feb 9, 2015 12:50:21 GMT
I cannot choose a single chromosome if I want to do a -MLMA-loco (leave one chr out).
MLMA analysis has no problem to work on >20chromosomes. The problem arises when I had the option "loco"
|
|
|
Post by SLacroix on Feb 9, 2015 13:11:15 GMT
In fact, I've just retested the --mlma option with --chr 21 and it says no SNP on chromosome 21, which isnt true.
The error message is: 0 SNPs on chromosome 21 are included in the analysis. Error: No SNP is retained for analysis.
same message for chromosome 21 through 26...
The --mlma-loco doesn't return any error message but stops at chr 20
Any idea?
Thanks
|
|
|
Post by Zhihong Zhu on Feb 17, 2015 6:13:38 GMT
Hi,
Maybe you can estimate allele frequency by GCTA or PLINK to check the SNPs on chr21-26. If no SNPs appear in those chr, please double check your plink file.
Cheers, Zhihong
|
|
|
Post by SLacroix on Feb 19, 2015 8:09:28 GMT
I have SNPs on those chromosomes.
I've done some more testing though and realized that the analysis stops in the middle of chr20. I've removed the SNP located where the analysis always stopped, reran it and end up with the same issue...
I've also built a test file comprised of randomly picked SNPs from chr 1 to 26. When the test file is not ordered by chr#, MLMA-loco with autosome-num 26 stops after the first SNP on chr26 it encounters...which is ok. If I order the test file by chr#, MLMA-loco with autosome-num 26 will stop somewhere in chr20...
Are you able to make some sense out of this?
|
|
|
Post by noa on Feb 20, 2015 22:32:01 GMT
Hi participants - I am unable to run --mlma I get "invalid option" error. Is it because I use command line on a PC (the way I do PLINK)? should I use Cygwin terminal ? thanks for ideas! - SLacroix - I also work on dog data. 38 autosomes. How did you end up with 26 ? just curious
|
|
|
Post by Zhihong Zhu on Feb 25, 2015 3:53:05 GMT
Hi Noa,
The current version of GCTA is complied on LINUX. Would you please try it in Cygwin terminal?
Cheers, Zhihong
|
|
|
Post by Zhihong Zhu on Feb 25, 2015 4:00:39 GMT
Hi SLacroix,
Sorry I have no idea about that. Would you please check those SNPs? eg. missing rate, reference allele frequency, or running association by R for those SNPs having problems. I'm sorry.
Cheers, Zhihong
|
|
|
Post by Jian Yang on Mar 17, 2015 2:13:44 GMT
I have tested this with a simple example. It actually works for data with number of chromosomes > 22.
gcta64_test --bfile test2 --mlma-loco --pheno test.phen --out test2 --autosome-num 25 --thread-num 10
******************************************************************* * Genome-wide Complex Trait Analysis (GCTA) * version 1.24.5 * (C) 2010-2013 Jian Yang, Hong Lee, Michael Goddard and Peter Visscher * The University of Queensland * MIT License ******************************************************************* Analysis started: Tue Mar 17 12:02:20 2015
Options: --bfile test2 --mlma-loco --pheno test.phen --out test2 --autosome-num 25 --thread-num 10
Note: the program will be running on 10 threads.
Reading PLINK FAM file from [test2.fam]. 3925 individuals to be included from [test2.fam]. Reading PLINK BIM file from [test2.bim]. 1000 SNPs to be included from [test2.bim]. Reading PLINK BED file from [test2.bed] in SNP-major format ... Genotype data for 3925 individuals and 1000 SNPs to be included from [test2.bed]. Reading phenotypes from [test.phen]. Non-missing phenotypes of 3925 individuals are included from [test.phen]. 3925 individuals are in common in these files.
Calculating allele frequencies ...
Calculating the genetic relationship matrix for each of the 25 chromosomes ... Chr 1: 40 SNPs on chromosome 1 are included in the analysis. Recoding genotypes (individual major mode) ...
Calculating the genetic relationship matrix (GRM) ... Chr 2: 40 SNPs on chromosome 2 are included in the analysis. Recoding genotypes (individual major mode) ...
......
Calculating the genetic relationship matrix (GRM) ... Chr 25: 40 SNPs on chromosome 25 are included in the analysis. Recoding genotypes (individual major mode) ...
Calculating the genetic relationship matrix (GRM) ...
Performing MLM association analyses (leave-one-chromosome-out) ...
----------------------------------- #Chr 1: 40 SNPs on chromosome 1 are included in the analysis.
Performing REML analysis ... (Note: may take hours depending on sample size). 3925 observations, 1 fixed effect(s), and 2 variance component(s)(including residual variance). Calculating prior values of variance components by EM-REML ... Updated prior values: 2.12466 3.78726 logL: -5315.06 Running AI-REML algorithm ... Iter. logL V(G) V(e) 1 -4950.22 0.92745 3.89896 2 -4890.78 0.65555 3.97793 3 -4878.34 0.52860 4.03307 4 -4873.65 0.45776 4.07170 5 -4871.64 0.41453 4.09880 6 -4870.73 0.38672 4.11784 7 -4870.29 0.32816 4.16018 8 -4869.87 0.32638 4.16295 9 -4869.87 0.32620 4.16305 10 -4869.87 0.32618 4.16306 Log-likelihood ratio converged.
Running association tests for 40 SNPs ... -----------------------------------
----------------------------------- #Chr 2: 40 SNPs on chromosome 2 are included in the analysis.
Performing REML analysis ... (Note: may take hours depending on sample size). 3925 observations, 1 fixed effect(s), and 2 variance component(s)(including residual variance). Prior values of variance components: 0.32618 4.16306 logL: -4747.16390 Running AI-REML algorithm ... Iter. logL V(G) V(e) 1 -4635.60 0.99411 3.27538 2 -4553.77 1.34882 3.05608 3 -4501.04 1.69213 2.91795 4 -4468.75 2.00819 2.82989 5 -4449.84 2.28084 2.77416 6 -4439.44 2.49732 2.74017 7 -4434.19 2.65302 2.72083 8 -4431.83 2.75458 2.71076 9 -4430.90 2.81635 2.70593 10 -4430.56 2.93126 2.69914 11 -4430.41 2.89365 2.70441 12 -4430.39 2.90694 2.70284 13 -4430.39 2.90225 2.70343 14 -4430.39 2.90391 2.70322 15 -4430.39 2.90333 2.70330 16 -4430.39 2.90353 2.70327 Log-likelihood ratio converged.
Running association tests for 40 SNPs ... -----------------------------------
......
----------------------------------- #Chr 25: 40 SNPs on chromosome 25 are included in the analysis.
Performing REML analysis ... (Note: may take hours depending on sample size). 3925 observations, 1 fixed effect(s), and 2 variance component(s)(including residual variance). Prior values of variance components: 2.75406 2.38225 logL: -4203.92111 Running AI-REML algorithm ... Iter. logL V(G) V(e) 1 -4203.86 2.73760 2.38978 2 -4203.84 2.73124 2.39561 3 -4203.83 2.73898 2.39473 4 -4203.83 2.73371 2.39534 5 -4203.83 2.73729 2.39493 6 -4203.83 2.73485 2.39521 7 -4203.83 2.73651 2.39502 8 -4203.83 2.73538 2.39515 9 -4203.83 2.73615 2.39506 10 -4203.83 2.73562 2.39512 11 -4203.83 2.73598 2.39508 Log-likelihood ratio converged.
Running association tests for 40 SNPs ... -----------------------------------
Saving the results of the mixed linear model association analyses of 1000 SNPs to [test2.loco.mlma] ...
Analysis finished: Tue Mar 17 12:11:18 2015 Computational time: 0:8:58
|
|