Using STARSolo SJ.out.tab?

[cnk113]

Hello,

I was wondering if it was possible to accept splice junctions generated from STAR/STARSolo?
I noticed it was slightly different from regtools in terms of junctions called. The data format is somewhat similar as well.
The format is outlined in the STAR manual.

Best,
Chang

[JBreunig]

I had no problem running Sierra with the STARsolo outputs. I'm curious to hear from the developer though.

[cnk113]

I just noticed in FindPeaks, it says SJ.out.tab is fine. I guess a more specific question I had multiple SJ.out.tab files from the same sample but different 10x lanes, it can be concatenated for the input?
Nvm, it seems to throw an error when I do so:

Import genomic features from the file as a GRanges object ... OK
Prepare the 'metadata' data frame ... OK
Make the TxDb object ... OK
36601 gene entries to process
Error in data.frame(..., check.names = FALSE) : 
  arguments imply differing number of rows: 2012124, 0
In addition: Warning message:
In .get_cds_IDX(mcols0$type, mcols0$phase) :
  The "phase" metadata column contains non-NA values for features of type stop_codon. This information was
  ignored.

[davhum]

Hi @cnk113 and @JBreunig

As you have discussed Sierra should work with STAR solo SJ.out.tab files. The FindPeaks function is designed to only take one SJ.out.tab file which should match the corresponding single BAM file from cellranger/STAR solo. If you concatenated splice junction files then multiple entries will occur for a number of splice junctions. This should not cause a crash but rather the read count for each duplicate would not be summed up. Therefore @cnk113 I am not sure I understand the source for the error you posted. Did this occur when you just passed one SJ.out.tab file?

If you did want to merge runs I would suggest running FindPeaks on each BAM file separately along with the corresponding SJ.out.tab file. Then you can merge peaks as described in the vignette before proceeding.

Regards,
David

[cnk113]

This error occurred after passing one SJ.out.tab file that was concatenated together. Also, in my case the separate runs have their own biases which I specifically do not want to normalize out, ideally the approach I want is the union of SJ.