Hi, I’ve just run crisprseq using the targeted pipeline with a read1.fastq.gz only. I heavily quality filtered the input reads, removing any reads containing N characters. In the output indels.csv, there are many cases of N characters being reported in the

pre_ins_nt

,

ins_nt

and

post_ins_nt

columns. When I check these reads in the input fastq file, the reported N characters are [ACGT] characters with Q value > 30. For the handful of reads I’ve looked at with these reported N characters, the majority called insertion (normal ACGT sequence) can be found in the input sequence, further suggesting these N calls could be erroneous results.

Hi @Alan Tracey, thanks for reporting this! I will have a look as soon as I get a moment. Would you mind opening an issue to the crisprseq repository? It would also be useful to have an example fastq file with some of these reads, for debugging.

Thank you Júlia. I will raise an issue, although I don’t think I will be able to share the fastq data as it is confidential. However, I have looked for examples of this in the test data set from the crisprseq github. Although they seem rare, there are cases of N’s reported in

M00724:1:000000000-DC7GJ:1:1102:13895:4077

which I can’t find in the

hCas9-AAVS1-a_R*

fastq files. Also, one other thing, when I run the test dataset using the ‘targeted’ pipeline with

-r 2.1.1

it works, but when I run it with the same command using

-r 2.2.0

it fails with:

-[nf-core/crisprseq] Pipeline completed with errors-
ERROR ~ Error executing process > 'NFCORE_CRISPRSEQ:CRISPRSEQ_TARGETED:ORIENT_REFERENCE (hCas9-TRAC-a)'

Caused by:
  The user value contains invalid characters. Enter a value that matches the pattern ^([a-z0-9_][a-z0-9_-]{0,30})$

Command executed:

  revComp_reference.R \
      hCas9-TRAC-a_reference.fasta \
      hCas9-TRAC-a_reference-correctOrient.fasta \
      ACAAAACTGTGCTAGACATG;
  
  cat <<-END_VERSIONS > versions.yml
  "NFCORE_CRISPRSEQ:CRISPRSEQ_TARGETED:ORIENT_REFERENCE":
      Rscript: $(Rscript --version)
  END_VERSIONS

The new version includes mostly changes to the

--analysis screening

part, only a couple of bug fixes from the

targeted

part, which shouldn’t affect your results.

This runs with no errors:

nextflow run nf-core/crisprseq -profile docker -params-file params.yaml -c awsbatch.config -r 2.1.1

....where params.yaml looks like:
input: './samplesheet_ko_only.csv'
outdir: 'output'
analysis: 'targeted'
overrepresented: True

….but changing to -r 2.2.0 gives the error above

It seems to be working for me. I’m not sure what throws this error as we don’t have anything similar in the pipeline. But it is complaining about a value which has to be shorter than 30 character and a combination of letters, numbers, dash or underscore. Are you providing any value which doesn’t match this pattern?

Ah, ok, thanks. I even get the error with the test dataset from the crisprseq pipeline, running with the files locally and names unchanged. But there is no error with -r 2.1.1

I found out this might have been introduced by a change on how we specify the docker user, should be fixed by https://github.com/nf-core/crisprseq/pull/163