Preparation

Setting up environment

Setting up conda environment

Install required packages in a new conda environment from environment.yml. If you are doing this on a local machine:

mamba env create -n patch_seq_spl

If you are doing this on Niagara, clone the repository to $SCRATCH and cd to the project root:

mamba env create -p ./envs/patch_seq_spl

• Use Python: Select Interpreter command in VS code to set the workspace-level Python interpreter as .env/bin/python
• Turn on python.terminal.activateEnvInCurrentTerminal
• Create a symbolic link, absolute path has to be used

ln -s /scratch/s/shreejoy/nxu/patch_seq_spl/env ${CONDA_PREFIX}/envs/patch_seq_spl

Setting up R environment

Since R packages on conda-forge channels are out-dated in general and the bioconda channel does not support osx-arm64, it's better to not install R packages within the conda environment.

Install R packages required for this project

remotes::install_github("nx10/httpgd")
install.packages(c("BiocManager", "tidyverse", "languageserver", "rlang", "arrow", "devtools", "reticulate", "svglite", "ggvenn", "rtracklayer"))
BiocManager::install(c("GenomicRanges", "GenomeInfoDb", "S4Vectors", "Gviz"))
devtools::install_github("dzhang32/ggtranscript")

Getting data

Download processed data from Niagara

rsync -av [email protected]:/scratch/s/shreejoy/nxu/patch_seq_spl/proc/ proc

Get genomic references

All genomic reference files can be found in $GENOMIC_DATA_DIR

cat > ~/.bashrc <<END
export GENOMIC_DATA_DIR="/project/s/shreejoy/Genomic_references/"
END

Get outputs from STAR

cp /scratch/s/shreejoy/nxu/patch_seq_spl/data/SJ_out_tabs/* data/SJ_out_tabs/

run Snakemake

Make sure that .env is activated. If you followed previous steps, .env should be automatically activated in any new terminals you create. Dry-run

snakemake --profile profile/default -np

Olego

Installation

1. Install pre-compiled Olego from https://sourceforge.net/projects/ngs-olego/files/
2. Create symbolic link to ~/miniforge3/condabin

ln -s /scratch/s/shreejoy/nxu/patch_seq_spl/tools/olego.bin.linux.x86_64.v1.1.5/olego /scratch/s/shreejoy/nxu/patch_seq_spl/env/bin/
ln -s /scratch/s/shreejoy/nxu/patch_seq_spl/tools/olego.bin.linux.x86_64.v1.1.5/olegoindex /scratch/s/shreejoy/nxu/patch_seq_spl/env/bin/

conda config --env --append channels conda-forge
conda config --env --append channels bioconda
mamba install --yes -c chaolinzhanglab quantas

git clone https://github.com/chaolinzhanglab/czplib
export PERL5LIB=/scratch/s/shreejoy/nxu/patch_seq_spl/tools/czplib

Build the index for the genome sequence

olegoindex /scratch/s/shreejoy/nxu/Genomic_references/mm39/Mus_musculus.GRCm39.dna.primary_assembly.fa -p proc/

Visualization

ggsashimi

Ran on dev02.camhres.ca:

python ~/tools/ggsashimi/ggsashimi.py -b proc/merge_bams/Sst_Calb2_Pdlim5.bam -c 2:66181571-66271179 -M 5 -g /external/rprshnas01/netdata_kcni/stlab/Nuo/patch_seq_spl/proc/Mus_musculus.GRCm39.110.gtf -o proc/figures/Sst_Calb2_Pdlim5 -F png --fix-y-scale --ann-height 3

However, the coordinates are covered by the annotation track. To get around this problem, I tried running ggsashimi in a docker container

Build docker image

singularity pull ggsashimi.sif docker://guigolab/ggsashimi:latest
singularity run -B $PWD:$PWD -W $PWD ggsashimi.sif -b proc/merge_bams/Sst_Calb2_Pdlim5.bam -c 2:66181571-66271179 -M 5 -g /external/rprshnas01/netdata_kcni/stlab/Nuo/patch_seq_spl/proc/Mus_musculus.GRCm39.110.gtf -o proc/figures/Sst_Calb2_Pdlim5_sashimi -F png

However, running ggsashimi gives out the error

1: In grDevices::png(..., res = dpi, units = "in") :
  unable to load shared object '/opt/R/3.6.3/lib/R/library/grDevices/libs//cairo.so':
  libtiff.so.5: cannot open shared object file: No such file or directory

Initially I thought this might be caused by the version of R in docker container overridden by conda R, but after commenting out the conda code in ~/.bashrc, the error persists.

gviz