Builder files

28 views
Skip to first unread message

Swethaa NG

unread,
Aug 26, 2024, 7:41:58 AMAug 26
to Biociphers
Hi,

Thank you for the excellent work you all do! I have run the commands and generated results, am currently in the analysing stage. While cross-checking with the pipeline steps, I came across some basic doubts about the input for build files which I would like to clarify.
 
Are there any specific parameters to be noted while STAR  mapping (the recommended tool, right?) that is suitable for known and novel junction recognition by MAJIQ? Like any thresholds or parameters to be set that makes the bam files more suitable for efficient splice graph formation?

Also, is the GENCODE .gff3 file okay? Or are there preferences for specific annotation files to be included? like from Ensembl or something?

Thank you!
Kind regards,
Swethaa

Benjamin Wales-McGrath

unread,
Aug 28, 2024, 4:33:56 PMAug 28
to Biociphers
Hi Swethaa,

In general, I would recommend standard ENCODE settings, and find these tend to work well. I would in general NOT recommend 2 pass mapping, as is sometimes done. This can uncover some additional splicing events, but these tend to be less reliable. Check out the ENCODE settings in the STAR manual section 3.2.2: https://physiology.med.cornell.edu/faculty/skrabanek/lab/angsd/lecture_notes/STARmanual.pdf

GENCODE GFF3 file is good!

Best regards,
Benjamin
Reply all
Reply to author
Forward
0 new messages