--tests option for chr 26

[ueki...@gmail.com]

Hi,

I found an inconsistency of --tests option for chr 26 between PLINKv1.90b4.2 and PLINKv1.07.


PLINKv1.90b4.2 .assoc.linear file:
 CHR     SNP                 BP   A1       TEST    NMISS       BETA         STAT            P
  26      MitoTC        200000    G        ADD          1000     -6.231       -2.519       0.0119
  26      MitoTC        200000    G        SEX          1000     0.5617        1.997      0.04608
  26      MitoTC        200000    G   USER_2DF     1000         NA           -9            0


PLINKv1.07 .assoc.linear file:
 CHR     SNP                 BP   A1       TEST    NMISS       BETA         STAT            P
  26      MitoTC        200000    G        ADD         1000     -6.231       -2.519       0.0119
  26      MitoTC        200000    G        SEX         1000     0.5617        1.997      0.04608
  26      MitoTC        200000    G   USER_2DF    1000         NA        9.987     0.006783


I think the line of STAT=-9 and P=0 from PLINKv1.90b4.2 is strange.


The .log file is as follows:

---------------------
PLINK v1.90b4.2 64-bit (7 May 2017)
Options in effect:
  --bfile test
  --chr 26
  --linear
  --noweb
  --out test
  --pheno pheno.txt
  --pheno-name age
  --sex
  --tests 1,2

Hostname: uuo
Working directory: /home/uuo
Start time: Wed May 10 09:06:05 2017

Note: --noweb has no effect since no web check is implemented yet.
Note: --sex flag deprecated.  Use e.g. '--linear sex'.
Random number seed: 1494374765
32109 MB RAM detected; reserving 16054 MB for main workspace.
135 out of 657366 variants loaded from .bim file.
1624 people (727 males, 897 females) loaded from .fam.
1252 phenotype values present after --pheno.
Using 1 thread (no multithreaded calculations invoked).
Before main variant filters, 1624 founders and 0 nonfounders present.
Calculating allele frequencies... done.
Total genotyping rate is 0.991001.
135 variants and 1624 people pass filters and QC.
Phenotype data is quantitative.
Writing linear model association results to test.assoc.linear ... done.

End time: Wed May 10 09:06:05 2017
---------------------


I am pleased if you could fix the inconsistency.

Thank,
Masao

[Christopher Chang]

Hi Masao,

Thanks for reporting this.  Can you send me a small dataset to reproduce this discrepancy with?

[ueki...@gmail.com]

Dear Chang,

Thanks for prompt reply.
I am attaching ped and map files.
The command is as follows.

plink --noweb --file test --linear --sex --out test1 --tests 1,2

Best wishes,
Masao

[Christopher Chang]

Thanks, I've posted a fix.

[ueki...@gmail.com]

Thanks, I checked that the bug is now fixed.

Best wishes,

Masao

2017年5月10日水曜日 14時49分19秒 UTC+9 Christopher Chang: