Repository navigation
riboWaltz with yeast data #92
Description
Activity
Hi there,
thanks for using RiboWaltz, and thanks for taking the time to go through the previous issues before opening a new one.I think the safest and most robust approach would still be to modify the GTF by adding the artificial UTRs there, then realign the reads. This way, everything downstream remains fully consistent, and you avoid having to patch annotations or intermediate objects afterwards, with the risk of introducing inconsistencies.
That said, if your only goal is to identify the P-site offsets, I think your approach of modifying
annotation_dtcould potentially be acceptable. The main issue I can think of, and which might explain the behavior you're observing, is related to the reads imported withbamtolist(). If I understood correctly, you only modified the annotation, so the 5' UTR and the transcript lengths are different from the original annotation (and I would also recommend extending the 3' UTR so that the total transcript length increases by 60 nt). However, the read coordinates (end5andend3) obtained from the BAM are still referenced to the transcript coordinates without the artificial UTRs.To keep everything consistent, you would also need to shift of 30 nt
end5andend3accordingly after importing the BAM files. Otherwise, the annotation and the read coordinates no longer refer to the same transcript coordinate system, which could explain why functions such asmetaprofile_psite()produce shifted profiles.I hope this helps, and please let me know if I've misunderstood any part of your workflow.
Best
Fabio
Hi,
I've been trying to run riboWaltz with yeast data with limited success. The issue as it has been discussed in this one and this one is the lack of UTR information in the GTF. I'm working with a custom GTF and Fasta files, generated by the lab I'm collaborating with. I have done alignment using STAR and producing "_Aligned.toTranscriptome.sorted" files. I have then created an annotation file as follows:
The problem with this annotation is that all UTRs are 0 and psite doesn't like that, but if I artificially I add:
psite function runs ok. However, the lack of 3UTR causes an issue with functions such
region_psite, as it selects transcripts with particular UTR length. I've tested altering the function to make l_utr3=0 and the function runs. My biggest issue is that when runningmetaprofile_psite, I get a metaprofile plot that is off centered and the signal in the 3' UTR is virtually gone (see file attached). I imagine is due to the artificially added UTR length, although I can't make full sense of it, as the peak is around nt, despite my artificial 5UTR value is 30 and "best offset: 12nts from the 5' end"My questions are, is my approach valid with regards to the 5'UTR? I'm only after the PO per read length to use elsewhere, or do I have to create a new gtf with UTRs added to it, realigned and rerun the analysis, as it was suggested in previous issues. Since my alignment is to the genome, despite generating a transcriptome outputs I would have thought the realignment is not required, but just double checking here. Thank you for your help!