Duplicate arc appears for each junction using example_run.sh

[britney-chu]

Hello,
I have recently cloned the most recent version of ggsashim (1.1.5) and created a conda environment with the env-r3.4.1.yml file. I then ran example_run.sh . It returned an error that I was missing pysam (I guess this doesn't come with the conda environment). After installing Pysam (v0.16.0.1) I ran example_run.sh again and it created a plot with two arcs for each junction, one above and one below.
my_ggsashimi_1_17.pdf

Any ideas of how to fix this?
Thanks!

[oliverking]

I likewise got duplicate arcs for the first plot in example_run.sh (the one with -O set but not -A), using both conda and the pre-built docker container. Based on a quick attempt to isolate the problem it seemed that the data.table aggregation commands that construct multi-sample labels for arcs in -O mode result in multiple rows for each arc since rows are not being collapsed along the "count" column when -A is not set (i.e., when "%(args.aggr)s" == ""):

ggsashimi/ggsashimi.py

Line 797 in a6d3c81

junctions = setNames(junctions[,.(max(y), max(yend),round(%(args.aggr)s(count)),paste(jlabel,collapse=",")), keyby=.(x,xend)], names(junctions))

ggsashimi/ggsashimi.py

Line 905 in a6d3c81

junctions = setNames(junctions[,.(max(y), max(yend),round(%(args.aggr)s(count)),paste(jlabel,collapse=",")), keyby=.(x,xend)], names(junctions))

I made an attempt to fix this issue in the fork at https://github.com/oliverking/ggsashimi, and this seems to work for the example plots, but I caution that I have not tested this change thoroughly, and have also not read though all the code, so don't know whether these changes might cause problems in other contexts.