Tool error n.2

[pietrisko]

Hi,

using the update version of the tool I get a permanent fail at the and, even if I have the output files from the deconvolution methods.

The command is:

cd mrna-prot
cwltool mrna-prot-comparison.cwl  isc-test2.yaml

Attaching package: ‘dplyr’

The following objects are masked from ‘package:stats’:

    filter, lag

The following objects are masked from ‘package:base’:

    intersect, setdiff, setequal, union

Combining 12 files
Error in read.table(file, fill = TRUE, check.names = FALSE) : 
  no lines available in input
Calls: main ... combinePatientCors -> do.call -> lapply -> FUN -> read.table
Execution halted
INFO [job get-patient-cors] Max memory used: 0MiB
ERROR [job get-patient-cors] Job error:
("Error collecting output for parameter 'table':\n../figures/plot-figs.cwl:36:9: Did not find output file with glob pattern: '['*.tsv']'", {})
WARNING [job get-patient-cors] completed permanentFail
ERROR [step get-patient-cors] Output is missing expected field file:///Users/pietro/Documents/GitHub/proteomicsTumorDeconv/perfEval/mrna-prot/mrna-prot-comparison.cwl#scatter-test/get-patient-cors/table
ERROR [step get-patient-cors] Output is missing expected field file:///Users/pietro/Documents/GitHub/proteomicsTumorDeconv/perfEval/mrna-prot/mrna-prot-comparison.cwl#scatter-test/get-patient-cors/fig
WARNING [step get-patient-cors] completed permanentFail
INFO [workflow ] completed permanentFail`

isc-test2.yaml is:

cancerTypes:
  - luad
  - brca
  - colon
prot-algorithms:
  - xcell
mrna-algorithms:
  - mcpcounter
  - epic
  - xcell
  - cibersort
tissueTypes:
   - tumor
signatures:
  - class: File
    path: ../../signature_matrices/LM22.txt

This does not happen if I use this yaml file, maybe the error is caused by mcpcounter and/or epic:

cancerTypes:
  - luad
  - brca
prot-algorithms:
  - xcell
mrna-algorithms:
  - xcell
  - cibersort
tissueTypes:
   - tumor
signatures:
  - class: File
    path: ../../signature_matrices/LM22.txt

Also: do you think it would be convenient to store the results for all tumors, signature matrixes and methods in a cloud or here? Because it takes a long time and computational power to run the tool and at this stage perhaps it is better to just download the results to do the subsequent analyses, maybe also adding the results in the final version of the tool

[LifeWorks]

Hi Pietro, did you remove your docker images and redownload them?

[pietrisko]

Hi Song, yes I did it.

[sgosline]

This is caused by #133 - if you ignore BayesDeBulk then it works fine.