Fix refactor_reportingtools_table.rb crash whenrun with non-ENS feature IDs and feature_type not 'gene'

[fischer-hub]

When we know we have valid ENS IDs we can use the features ENS stable ID prefix (ENSG, ENST etc.) to scan for the gene ID in the HTML report from reporting tools. This fails with arbitrary IDs if the feature the GTF was filtered for is not also the gene ID because we then have a gene ID from the reporting tools report and an exon / transcript ID from the filtered GTF that we cant map to each other.

• added valid ENS ID check
• added additional mapping from gene ID to feature ID from the GTF file to later map to the feature name, gene biotype and features position in case non-ENS IDs were detected and standard mapping fails (feature type is not 'gene')

closes #204

[hoelzer]

@fischer-hub looks good now, great! I now get the correct PCAs on this branch (do we still needed then PR #209 ?)

Two things:

1. do you know why in the ReportingTools output the images are not shown? E.g.:

2. Do you know why this folder _vs_SBY is generated:

(base) ➜  rnaflow-test-data-david ls results-SBY92ref-CBS14_P2_vs_SBY92_P3-debugging/07-DifferentialExpression/DESeq2/                                                                  
CBS14_P2_vs_SBY92_P3  data  deseq2.Rout  plots  _vs_SBY

The content is just:

results-SBY92ref-CBS14_P2_vs_SBY92_P3-debugging/07-DifferentialExpression/DESeq2/_vs_SBY
└── downstream_analysis
    └── piano

2 directories, 0 files

so maybe this is generated during the piano function?

[fischer-hub]

@hoelzer No, I think I'll delete #209

1. do you know why in the ReportingTools output the images are not shown? E.g.:

I think you are looking at the raw Reportingtools report no? The refactored one has 'extended' as a suffix. The extended report for me shows the images.

2. Do you know why this folder `_vs_SBY` is generated:

Nice catch, yes there was a broken regex in the Webgestalt and Piano processes that didn't match the right comparison names :D I added a fix in the last commit.

[hoelzer]

Yes then let's simply delete PR #209 !

Ah, I thought the "normal" report also had these little thumbnails - okay anyway you're right!

I tested the branch again and the folder is gone now!

Looks all good from my side. I would do a merge and little release w/ these fixes for non-model input annotations. Thx!