4.reports.pbs crashing

[ivan108]

No description provided.

[ivan108]

I have a lot of problems with the last step of the pipeline.
After steps 1 - 3 successfully done, the 4.reports.pbs script fails with the reproducible error:

Sample 1 ('Patient157t10,Z00600t10_vs_Z00599t10') of 2 ...
...done
Sample 2 ('Patient157t10,Z00601t10_vs_Z00599t10') of 2 ...
[...]
Sample 2 ('Patient157t10,Z00601t10_vs_Z00599t10') of 2 ...
...done
Error in texi2dvi(pathnameR, pdf = pdf, clean = clean, quiet = quiet,  :
  Running 'texi2dvi' on 'Patient157t10,Z00600t10_vs_Z00599t10,PairedPSCBS,report.tex' failed.
BibTeX errors:
I couldn't open database file PSCBS.bib
---line 61 of file Patient157t10,Z00600t10_vs_Z00599t10,PairedPSCBS,report.aux
 : \bibdata{bioinformatics-journals-abbr,PSCBS    

Here is the output in "reports" directory:

Costello-PSCN-Seq/reports/Costello-Exome,bwa,realigned,rmDups,recal,seqz,100kb,tcn=2: ls
bioinformatics-journals-abbr.bib
figures
includes
natbib.bst
Patient157t10,Z00600t10_vs_Z00599t10,PairedPSCBS,report.aux
Patient157t10,Z00600t10_vs_Z00599t10,PairedPSCBS,report.bbl
Patient157t10,Z00600t10_vs_Z00599t10,PairedPSCBS,report.blg
Patient157t10,Z00600t10_vs_Z00599t10,PairedPSCBS,report.log
Patient157t10,Z00600t10_vs_Z00599t10,PairedPSCBS,report.pdf
Patient157t10,Z00600t10_vs_Z00599t10,PairedPSCBS,report.tex
Patient157t10,Z00600t10_vs_Z00599t10,PairedPSCBS,report.tex.rsp
Patient157t10,Z00600t10_vs_Z00599t10,PairedPSCBS,report.toc
Patient157t10,Z00601t10_vs_Z00599t10,PairedPSCBS,report.tex.rsp
reports

So, it seems that the report for one sample is successfully generated, but fails for the second one?

The most weird thing is that sometimes it actually works!?!?

Any ideas?
Thx!

[ivan108]

Today I got a different error message:

Sample 1 ('Patient157t10,Z00600t10_vs_Z00599t10') of 2 ...
...done
Sample 2 ('Patient157t10,Z00601t10_vs_Z00599t10') of 2 ...
...
Sample 2 ('Patient157t10,Z00601t10_vs_Z00599t10') of 2 ...
...done
Error in createLink.default(link = link, target = target) :
  Failed to create file link: reports/Costello-Exome,bwa,realigned,rmDups,recal,seqz,100kb,tcn=2/PSCBS.bib[.lnk] -> /home/ismirnov/R/x86_64-pc-linux-gnu-library/3.4/PSCBS/templates/rsp/PSCBS.bib
Calls: unlist ... eval -> eval -> value -> value.Future -> resignalCondition

I checked, the file /home/ismirnov/R/x86_64-pc-linux-gnu-library/3.4/PSCBS/templates/rsp/PSCBS.bib does exist, what could prevent making a link to it??

[HenrikBengtsson]

Sorry, this one slipped. Do you still have this problem? It's not clear to me from your report whether it ever worked for you or not - is/was this a problem that appeared out of nowhere?

If you launch the same version of R (looks like R v3.4.?), does it work when you do:

library(PSCBS)
J <- 1000
mu <- double(J)
mu[200:300] <- mu[200:300] + 1
mu[350:400] <- NA # centromere
mu[650:800] <- mu[650:800] - 1
y <- mu + rnorm(J, sd=1/2)
fit <- segmentByCBS(y)
pdf <- report(fit, sampleName = "foo", studyName = "bar")
print(pdf)

[ivan108]

Thanks for getting back to me. I could use your help to make it working. Temporary I gave up on it, went on vacation for big part of January. Currently working on Mutect2 implementation using the latest GATK4.1 and some other projects.
Will go back to PSCN as soon as I get Mutect2 working...

As for the code, it works if I load R 3.4.1 or 3.4.4

[HenrikBengtsson]

FYI, I just ran through the pipeline and I did not have any issues with the report generation. I updated to make use of module load r/3.5.2 but it should work the same with r/3.4.4. Maybe it was a hiccup?

[ivan108]

I tried to run the PSCN pipeline using LG3 interface on the weekend using small test dataset, but it didn't produce reports.
But today I ran it 3 times, and it did work every time! I guess it was a hiccup...

[HenrikBengtsson]

Interesting. Maybe it related to which node the jobs were running on. Also, does "it didn't produce reports" mean the originally reported error, or something different?

[ivan108]

What happens is that the first 3 steps (mpileup, sequenza and pscbs) produce output which looks right, but the reports step quietly dies without any errors and output. On some other occasions reports produces partial output...