Extra column Kinship matrix

[medinacarolina]

Dear Xiaowei, for the GIANT/GLC project I have been running all my analysis using rvtest Program version: 20170307. Somehow all of them produce the same log file where an error is reported "[ERROR] Unexpected column 'u5401'!". For all the 3 cohorts I am managing

But then after the program seems to continue normally. Adam Locke suggested the problem is just affecting chrX, so that I should submit all the autosomals results and communicate with you before submitting chrX results (more than 60 GWAS have already ran). Thus, I need to know my results are valid, and where the problem came from.

I am attaching the log from the kinship creation, an example for the autosomals log file and one for chrX. Please confirm the validity of the files.

Thanks a lot. Carolina RS2_CEU_1000G_imputation_TC_INV_MALE_chr17_log.txt RS2_CEU_1000G_imputation_TC_INV_MALE_chrX_log.txt RS2_kinship_matrix.vcf2kinship_log.txt

[zhanxw]

Hi Carolina,

Usually you can safely ignore this warning. Let me check once more to make sure of that.

Thanks. Xiaowei

[medinacarolina]

Thanks Xiaowei, pleae double check and let me know... also, concerning the chrX analysis in the log I see not mention to the loading of the xHemiKinship matrix, is this normal (I am attaching again an example of the log for chrX analysis)? RS2_CEU_1000G_imputation_TC_INV_MALE_chrX_log.txt

[tesselgalesloot]

Hi Carolina and Xiaowei, I have exactly the same issue with regard to the error of the unexpected column. Could you indeed confirm that we can ignore this error? Regarding the chrX analysis the loading of the kinship matrices is mentioned in my logs, but I specified both the autosomal and X kinship matrix in the command line, also for X analysis, so that can be a difference. Thanks in advance, Tessel

[zishang30]

Dear Xiaowei

I have started running chrX analysis using generated kinship and xHemi Kinship. But when I running the command as follow : rvtest-new --inVcf ChrX.combined.reorder.vcf.gz --pheno phenotypes0602.ped --pheno-name men_INV --kinship kinship_matrix.kinship --xLabel X --xHemiKinship chrx.kinship_matrix.xHemi.kinship --meta score,cov[windowSize=500000] --dosage DS --out HRCinv_men_chr23

I found some strange message as follows: [INFO] Loaded 1386 male, 1490 female and 0 sex-unknonw samples from phenotypes.ped [INFO] Analysis begins with [ 2876 ] samples... [INFO] Family-based model specified. Loading kinship file... [INFO] DONE: Loaded kinship file [ kinship_matrix.kinship ] successfully in [ 11.0 ] seconds. [INFO] DONE: Spectral decomposition of the kinship matrix succeeded in [ 124.7 ] seconds. [INFO] Kinship eigen-decomposition file [ .xHemi.eigen ] detected and will be used for for X chromosome analysis [INFO] DONE: Loaded kinship file [ chrx.kinship_matrix.xHemi.kinship ] successfully in [ 20.9 ] seconds. [ERROR] Missing 'u1805' column! [WARN] Failed to load spectral decomposition results of the kinship matrix! [INFO] DONE: Spectral decomposition of the kinship matrix succeeded in [ 202.1 ] seconds. [INFO] Impute missing genotype to mean (by default) [INFO] Use dosage genotype from VCF flag DS. [INFO] Analysis started

It seems there are some error about missing column in kinship matrix and lead to fail to load spectral decomposition results of the kinship matrix, but after this warning message it said "Spectral decomposition of the kinship matrix succeeded". Is this Ok to run chrX analysis ? The strange thing is when I check chrx kinship matrix manually I didn't find this missing column in 'u1805'...

In addition , some times I also receive error massage about 'Unexpected column' as the previous issue...

Thank you for your help!

zishan

[zhanxw]

Hi Zishan,

Can you please check how many columns in the chrx kinship matrix file? Thanks.

[zishang30]

Dear Xiaowei

I have checked chrx kinship matrix file, it has 4150 columns compared with 4149 rows. But this could be possible because the format of kinship matrix is as follow: FID IID ID1 ID2 ID1 ID2 value1 value2

compare to rows, it has extra column of IID.

I plan to start finish the left chrx analysis at today. Is this a problem?

Thanks.

[zhanxw]

The extra column of IID may be the reason of 'Extra column Kinship matrix'.