New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Coxph interactions not working #15

Closed
lckarssen opened this Issue Jan 12, 2016 · 2 comments

Comments

Projects
None yet
1 participant
@lckarssen
Member

lckarssen commented Jan 12, 2016

This is a 'liftover' of bug #6194 in the R-forge tracker.

Reported on ProbABEL v0.4.5.

See forum post http://forum.genabel.org/viewtopic.php?f=10&t=975. From there:

When our phenotype file includes four columns (ID, Age, Event, cov1) and the interaction parameter is specified as interaction==1, the model given is incorrect (Coxph model: ( Age , Event ) ~ mu + cov1 + SNP_A1 + mu*SNP_A1).

When the interaction parameter is specified as interaction==2, the model shown is now correct, Coxph model: ( Age , Event ) ~ mu + cov1 + SNP_A1 + cov1*SNP_A1, but the analysis doesn't run, giving the following error
error: Interaction parameter is out of range (interaction=2)

When we include a phenotype file with an additional covariate column (ID, Age, Event, cov1, cov2) and specify interaction==2, the model looks ok (with the exception of a covariate we didn't want to include), Coxph model: ( Age , Event ) ~ mu + cov1 + cov2 + SNP_A1 + cov1*SNP_A1, and the analysis seems to run, but it's not clear what analysis is being run to give the results; we can't replicate any of the results shown in R.

@lckarssen lckarssen added this to the v0.5.0 milestone Jan 12, 2016

@lckarssen lckarssen modified the milestones: v1.0.0, v0.5.0 Feb 8, 2016

@lckarssen lckarssen self-assigned this Apr 22, 2016

@lckarssen

This comment has been minimized.

Show comment
Hide comment
@lckarssen

lckarssen Apr 22, 2016

Member

I have just tested this on commit 4180862 and this bug is indeed present.

Member

lckarssen commented Apr 22, 2016

I have just tested this on commit 4180862 and this bug is indeed present.

lckarssen added a commit to lckarssen/ProbABEL that referenced this issue Apr 22, 2016

Fix Issue #15: interactions in Cox broken
This fixes the --interactions option when using pacoxph. With the
introduction of the (mathematically useless) \mu in the Cox model (which
was introduced to fix bug 1266 (pacoxph with no covariates) on R-forge,
https://r-forge.r-project.org/tracker/index.php?func=detail&aid=1266&group_id=505&atid=2058),
the extra parts of code where we adjusted for the missing \mu were
inadvertently kept. Cox + interactions must have been broken since then.

lckarssen added a commit that referenced this issue Apr 22, 2016

@lckarssen

This comment has been minimized.

Show comment
Hide comment
@lckarssen

lckarssen Apr 22, 2016

Member

Closed by PR #26.

Member

lckarssen commented Apr 22, 2016

Closed by PR #26.

@lckarssen lckarssen closed this Apr 22, 2016

@lckarssen lckarssen added the pacoxph label Nov 22, 2016

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment