COXPH and matrix errors

Questions about ProbABEL are welcome here.
Forum rules
Please remember not to post any sensitive data on this public forum.
The first few posts of newly registered users will be moderated in order to filter out any spammers.

When get a solution to the problem you posted, please change the topic name (e.g. from "how to ..." to "[SOLVED] how to ..."). This will make it easier for the community to follow the posts yet to be attended.
hkins552
Posts: 2
Joined: Wed Jun 10, 2015 5:41 pm

COXPH and matrix errors

Postby hkins552 » Wed Jun 10, 2015 7:04 pm

Hi,
I'm a new Probabel user attempting to trial this software for an upcoming GWAS. During the run, every SNP seems to have a matrix error:

Code: Select all

Warning for rs724512: X matrix deemed to be singular (variable 2)
Warning for rs1188382: X matrix deemed to be singular (variable 2)
Warning for rs1188383: X matrix deemed to be singular (variable 2)
Warning for rs1188384: X matrix deemed to be singular (variable 2)
Warning for rs111627626: X matrix deemed to be singular (variable 2)
Warning for rs113280400: X matrix deemed to be singular (variable 2)
Warning for rs1188385: X matrix deemed to be singular (variable 2)
Warning for rs77506101: X matrix deemed to be singular (variable 2)
Warning for rs2147081: nr of iterations >= MAXITER (20): 20
Warning for rs2147081: Cox regression ran out of iterations and did not converge, setting beta and se to 'NaN'
Warning for rs1188386: X matrix deemed to be singular (variable 2)
Warning for rs79951922: X matrix deemed to be singular (variable 2)
Warning for rs76912545: X matrix deemed to be singular (variable 2)
Warning for rs1188387: X matrix deemed to be singular (variable 2)
Warning for 131149813TCT: X matrix deemed to be singular (variable 2)
Warning for rs1188388: X matrix deemed to be singular (variable 2)
Warning for rs1188389: X matrix deemed to be singular (variable 2)
Warning for rs75262501: X matrix deemed to be singular (variable 2)
Warning for rs78715304: X matrix deemed to be singular (variable 2)
Warning for rs1920423: X matrix deemed to be singular (variable 2)
Warning for rs2700541: X matrix deemed to be singular (variable 2)
Warning for rs1408151: X matrix deemed to be singular (variable 2)
Warning for rs1188390: X matrix deemed to be singular (variable 2)


The first 5 lines of my info:

Code: Select all

#SNP   Al1   Al2   Freq1   MAF   Quality   Rsq
rs116559375   A   G   30028088   0.5   0.9   0.9
rs116379285   C   T   30028139   0.5   0.9   0.9
rs116095946   T   C   30028141   0.5   0.9   0.9
rs74066615   T   C   30028236   0.5   0.9   0.9
rs72652657   C   G   30028329   0.5   0.9   0.9
rs67913534   T   C   30028437   0.5   0.9   0.9
rs28742271   T   C   30028717   0.5   0.9   0.9
rs28588071   G   A   30028730   0.5   0.9   0.9
rs78986157   T   C   30028825   0.5   0.9   0.9

(With final three columns arbitrarily set as this info was not provided in our original dosage files)

Dose file:

Code: Select all

TOR1201-04 MLDOSE 0.331 0.327 0.064 0.139 0.081 1.533 0.328 1.191
TOR1203-03 MLDOSE 0.765 0.766 0.14 0.196 0.213 1.038 0.765 0.069
TOR1216-03 MLDOSE 0.855 0.855 0.368 0.406 0.268 0.738 0.857 0.109
TOR1228-03 MLDOSE 0.183 0.183 0.013 0.02 1.738 1.797 0.183 0.028
TOR1244-03 MLDOSE 0.871 0.869 0.105 0.27 0.224 0.86 0.869 0.249
TOR1255-03 MLDOSE 1.192 1.191 0.05 0.112 0.161 0.697 1.194 0.111
TOR1261-03 MLDOSE 1.091 1.091 0.145 0.163 0.038 0.746 1.091 0.6
TOR1261-04 MLDOSE 0.411 0.405 0.093 0.17 0.342 1.427 0.403 0.863
TOR1265-03 MLDOSE 0.91 0.91 0.01 0.015 0.009 1.074 0.911 0.041
TOR1265-04 MLDOSE 0.636 0.636 0.134 0.155 0.68 1.21 0.634 0.297

Pheno file:

Code: Select all

id fupt_chd chd   arbitrarycov
TOR1201-04 9.9150684932 0   1
TOR1203-03 11.860273973 0   1
TOR1216-03 9.2712328767 0   1
TOR1228-03 14.043835616 0   1
TOR1244-03 11.010958904 0   1
TOR1255-03 11.068493151 0   1
TOR1261-03 12.830136986 0   1
TOR1261-04 10.8 0   1
TOR1265-03 10.052054795 0   1

(Had to add last column -fake covariate- to achieve run)

In total, I have ~40,000 snps and ~300 samples in this dataset.
Any suggestions? I apologize if I could not find the answer in previous posts.
Last edited by lckarssen on Fri Jun 12, 2015 8:18 am, edited 1 time in total.
Reason: Reformatted part of the input and output with the [code] tag

lckarssen
Site Admin
Site Admin
Posts: 321
Joined: Tue Jan 04, 2011 3:04 pm
Location: Utrecht, The Netherlands

Re: COXPH and matrix errors

Postby lckarssen » Fri Jun 12, 2015 7:56 am

Which version of ProbABEL are you using? We recently released a bug fix (v0.4.5) that fixed a part of the code that erroneously generated these errors.
-------
Lennart Karssen
PolyOmica
The Netherlands
-------

lckarssen
Site Admin
Site Admin
Posts: 321
Joined: Tue Jan 04, 2011 3:04 pm
Location: Utrecht, The Netherlands

Re: COXPH and matrix errors

Postby lckarssen » Fri Jun 12, 2015 8:20 am

By the way:
hkins552 wrote:Pheno file:

Code: Select all

id fupt_chd chd   arbitrarycov
TOR1201-04 9.9150684932 0   1
TOR1203-03 11.860273973 0   1
TOR1216-03 9.2712328767 0   1
TOR1228-03 14.043835616 0   1
TOR1244-03 11.010958904 0   1
TOR1255-03 11.068493151 0   1
TOR1261-03 12.830136986 0   1
TOR1261-04 10.8 0   1
TOR1265-03 10.052054795 0   1

(Had to add last column -fake covariate- to achieve run)

This last remark leads me to believe you aren't yet using the latest ProbABEL release, because that bug is fixed in 0.4.5 as well.
-------
Lennart Karssen
PolyOmica
The Netherlands
-------

hkins552
Posts: 2
Joined: Wed Jun 10, 2015 5:41 pm

Re: COXPH and matrix errors

Postby hkins552 » Fri Jun 12, 2015 2:11 pm

I actually am using the latest version:
/usr/local/bin/probabel-0.4.5/src/pacoxph

lckarssen
Site Admin
Site Admin
Posts: 321
Joined: Tue Jan 04, 2011 3:04 pm
Location: Utrecht, The Netherlands

Re: COXPH and matrix errors

Postby lckarssen » Tue Jun 16, 2015 12:04 pm

hkins552 wrote:I actually am using the latest version:
/usr/local/bin/probabel-0.4.5/src/pacoxph

Hmm, that's unexpected 8-).

Looking at the first output you posted, I see that variable 2 gives you the singular matrix error. This is the dummy covariate you added. The fact that it's constant triggers the message.
Could you try again without the covariate. ProbABEL 0.4.5 should be able to process such input. If not, please post the output of the pacoxph command you're using.
-------
Lennart Karssen
PolyOmica
The Netherlands
-------


Return to “ProbABEL”

Who is online

Users browsing this forum: Bing [Bot] and 1 guest