suppa

[leipzig]

Reviewer comments
for fig 1, we use SUPPA2 to identify annotated splicing. it takes a GTF in and outputs a format call IOE. for splicing, we generally avoid calling the event at the script level. https://github.com/comprna/SUPPA
XMA Header Image
comprna/SUPPA
github.com

Eric sent Today at 12:21 PM
we write our own algorithm to call novel/unannotated events because we haven't found one that optimizes sensitivity/specificity as well as our internal development
You sent Today at 12:22 PM
oh great

Eric sent Today at 12:22 PM
once you have the events, calculating PSI is straightforward
you can either use deseq/dexseq type of workflow to call different spliced junctions with raw count expression between your cohorts, or whatever statistics that fancy you to do differential using PSI

and for conjoined, id just use any fusion algorithm (like star-fusion for instance), then parse the output such that both genes of the fusion product are located in the same chromosome

import os
from snakemake import shell

gtf = os.path.splitext(snakemake.input.gtf)[0]
ioe = os.path.splitext(snakemake.output.ioe)[0]

shell(
"""
pigz -d -c {snakemake.input.gtf} > {gtf}
suppa.py generateEvents -i {gtf} -o {ioe} -e {snakemake.params.events} -f ioe
rm -f {gtf}
""")

merge all the ioe files into the final output

shell(
"""
awk '
FNR==1 && NR!=1 {{ while (/^seqname/) getline; }}
1 {{print}}
' {ioe}*.ioe > {snakemake.output.ioe}

rm {ioe}*strict.ioe {ioe}*strict.gtf
""")

rule run_suppa_generate_events:
"""
Run suppa to generate alternative splicing events.
"""
input:
gtf = '_commons/analysis/references/{ref_source}.{ref_species}.{ref_version}.comprehensive.chr.gtf.gz'
output:
ioe = '_commons/analysis/references/{ref_source}.{ref_species}.{ref_version}.comprehensive.chr.gtf.ioe'
params:
events = ['SE', 'SS', 'MX', 'RI', 'FL']
conda:
'../../../conda_envs/bio/suppa/2.3.yml'
script:
'scripts/suppa/generate_events.py'

try suppa2 and see if you can replicate figure 1 first. also, instead of fold change, we usually use delta PSI in splicing, similar to what the paper uses