Search the entire project
This project's trackers
This project's news
[#1266] pacoxph with no covariates
Yurii Aulchenko (
Lennart Karssen (lckarssen)
pacoxph with no covariates
From Aron Joon:
We tried to run Cox models using pacoxph
An error message will come up as "nc <=0" when I tried to model "phenotype : SNPs" (ie, no covariate). Is there a trick how to fix this? I added a constant dummy (covariate) variable (all values = 1) in order to get ProbABEL run smoothly
Date: 2015-03-17 13:55
Fixed in r1931.
Date: 2015-03-06 10:50
See also Bug #6041 for more (possibly related?) info.
Date: 2015-01-25 17:32
According to one user report sent to me by e-mail, as of ProbABEL v0.4.4, the dummy covariate work-around doesn't work any more, but leads to the error message "X matrix deemed to be singular, setting beta and se to 'NaN'" for all SNPs.
With a non-constant covariate (e.g. sex), the output is ok.
Date: 2014-05-26 14:59
I added a check for this bug in r1740.
Currently the check is listed as XFAIL (expected FAIL) in Makefile.am. Once this bug is fixed, 'make check' should give an error (and the check should be listed as XPASS). That's the time to remove the XFAIL line from Makefile.am and to extend the check to include a diff with verified data.
Date: 2014-03-16 12:01
Correction to the previous message, the SVN revision number was r1644 (this is the revision in which I slightly edited the error message).
Date: 2014-03-16 11:58
I just ran into this bug (PA SVN r1631).
The "nc <- 0" message appears when ProbABEL is not compiled with EIGEN. If EIGEN is used, then the message is:
mematrix(): number of columns less than 1
Accepted As Bug