A community effort to collect a curated set of analysis pipelines built using Nextflow.

nf-core

Hi everyone,
I have been implementing the nf-gpt plugin into nf-core/crisprseq. The plugin generates text which I am writing into a file (not through a module). Is it possible to incorporate the generated file, from `collectFile()` function, into the pipelines multiqc report?
<https://github.com/nf-core/crisprseq/pull/193>
I am new to multiqc so I'd appreciate any help you can give me :slightly_smiling_face:

Yes, it should be! MultiQC can take basically any inputs. 

The simplest is probably to use Custom Content - just create a file and end it in _mqc.txt (or whatever file format)

See <https://docs.seqera.io/multiqc/custom_content|https://docs.seqera.io/multiqc/custom_content>

Note that MultiQC already has AI summaries built in, but not sure what it is that you're building :) <https://docs.seqera.io/multiqc/ai|https://docs.seqera.io/multiqc/ai>

Amazing, I will have a look at it. I am aware that MultiQC has AI integration already. From what I can see my implementation tackles a slightly different angle and is less focused on summarising. It is more of an experimental (very optional) implementation that is communicated with the core pipeline maintainers.

gpt_drugz_output_mqc.txt

Hi <@UE4L64T4G> sorry for the ping. I am only just getting to work on this and have had the chance to look into the documentations.

From what I can see the custom content works for .txt files. But it also looks like it is designed for tabular data specifically. Is this correct? I am trying to incorporate a text file with plain gpt generated text (see example file).  Meaning I currently cannot reliably enforce any sort of structure for the output.

I tried rust running `multiqc .` in the files directory and it gives me `Inconsistent number of columns found in gpt_drugz_output_mqc.txt.` . Do you have any idea how the file content could still be included? Or maybe a workaround?

It works with many file extensions. From the docs:
&gt; `*_mqc.(yaml|yml|json|txt|csv|tsv|log|out|png|jpg|jpeg|html)`

It then tries to guess what it should do with that file based on the file contents. Or any config that you supply.

I was about to suggest that you use `.md` and give it markdown (as it looks like that's what you have). But actually I'm not sure that we support that out of the box.. So that would be a great feature request github issue! It'd be super easy for us to implement.

Anyway, for right now - I think you are best off using HTML. For example (from the <https://docs.seqera.io/multiqc/custom_content|docs>):

```&lt;!--
id: 'custom-html'
section_name: 'Custom HTML'
description: 'This section is created using a custom HTML file'
--&gt;
&lt;p&gt;Some custom HTML content here.&lt;/p&gt;```

Then ask your LLM to give you HTML instead of markdown

Made an issue for supporting markdown: <https://github.com/MultiQC/MultiQC/issues/3199>

Thank you, i'll see if I can get it running :clap: