Magdoll / SQANTI2

SQANTI2 is now replaced by SQANTI3. Please go to: https://github.com/ConesaLab/SQANTI3
Other
38 stars 15 forks source link

Tamma Collapse for 5' cap selected samples #57

Open BeatrizdeToledo opened 4 years ago

BeatrizdeToledo commented 4 years ago

Good afternoon, For my PAC-BIO sequencing I performed the 5' cap selection step. I recently found out that the Iso-Seq Tofu algorithm for collapsing is not appropriate for data that performed 5' cap selection step.

On the other hand, the TAMA Collapse is designed for TAMA Collapse is designed for 5' cap library as well. https://github.com/GenomeRIK/tama/wiki/Tama-Collapse In this link they explain a little more with tofu is not appropriate for 5' selected samples.

the cDNA_Cupcake used in SQANTI2 behaves as TAMA Collapse? https://github.com/Magdoll/cDNA_Cupcake/wiki#install Is it possible to add the possibility to use TAMA Collapse to Sqanti2 script, or to make cDNA_cupcake suitable to 5'selected samples? Or is it better to do the mapping and the collapsing step separetly, and only after the TAMA Collapse step start to use the sqanti2 script?

Magdoll commented 4 years ago

Hi @BeatrizdeToledo ,

You can run Cupcake collapse collapse_isoforms_by_sam.py with the option --dun-merge-5-shorter which will preserve 5' differences at the exonic level but will still collapse isoforms with the same 5' exon but start site may vary slightly within that exon. If you want full control, you probably should go with TAMA collapse.

As long as you can create a GFF3 file out of the TAMA collapse output, SQANTI2 can be run.

It does look like there are some ways to convert BED to GFF3. If this becomes a persistent issue, let me know and send me a test BED file to work with and I can see if I can make it SQANTI2 compatible.