First call src/suppa2_generate_events.sh
to create the ioe files with events descriptions. Then, as indicated in a comment in src/suppa2_psiPerEvent.sh
, group these ioe for each event type in a single one.
Separately, bring in quantifications from stringtie (see stringtie_quantifs
repo). If needed, prefilter by removing outliers with R/str_q_remove_outliers.R
Run src/suppa2_psiPerEvent.sh
.
Run src/suppa2_dpsi.sh
, which will call src/split_events.R
to group PSI by neuron type.
Main analysis in R/dpsi.R
. This script uses accessory functions defined in R/sequence_properties.R
, R/sequence_conservation.R
, and R/extract_event_coordinates.R
.
Additional exploratory and tests scripts are stored in the branch exploratory
.
I initially thought about using Salmon for quantifications: the fact that we need to deduplicate UMIs makes it non-trivial, deleted (can be found in old commits).