single-cell RNA Isoform Analysis Pipeline
Author: Vivek Bhardwaj (@vivekbhr) and Thijs Makaske
From Conda:
conda create -n scria -c vivekbhr -c bioconda -c conda-forge scriapipe
From GitHub:
conda create -n scria python pip
conda activate scria && pip install snakemake
pip install git+https://github.com/vivekbhr/scRIApipe.git@master
The semi-permanent parameters to the workflow could be configured with a yaml file (config.yaml). Download here
After conda install, move to the folder where you wan to run the workflow, and prepare the config.yaml
cd <your output dir>
vim config.yaml ## now modify the paths as per your requirements
The workflow needs
- path to a GTF file (preferably pre-filter the GTF to remove pseudogene and low confidence annotations)
- UCSC ID of the genome
cDNA fasta and GTF can be downloaded here UCSC ID is, for example "mm10" (mouse) or "hg38" (human)
Additional parameters for the workflow can be accessed via --help
scRIA --help
## inside the output dir
scRIA -i <fastq_folder> -o . -c <your>config.yaml -j <jobs> -s ' -np'
-
In the workflow command above, j is the number of parallel jobs you want to run, -cl means submit to cluster (default is to run locally). Therefore if you wish to run the workflow on a cluster, simply use the workflow with the -cl command on the submission node.
-
cluster configuration, such as memory and cluster submission command are placed in cluster_config.yaml, and can be modified to suite the users internal infrastructure.
To find the currently active cluster-config file, run the following:
conda activate scria
which scRIA
This shows the path to scRIA binary, usually it's like:
/path/to/miniconda3/envs/scria/bin/scRIA
You can find the cluster-config here:
/path/to/miniconda3/envs/scria/lib/python<version>/site-packages/scriapipe/cluster_config.yaml
In order to just test what the workflow would do, use the command -s ' -np'
Index builing needs >40G of memory, if the workflow fails and the logs/velocity_index.err says something like std::badalloc, increase memory in the file cluster_config.yaml in the scRIA folder.
-
After running the pipeline, LOG file are stored in the /log/ directory and the workflow top-level log is in scRIA.log file.
-
Currently the -o option is not very flexible and and pipeline works only when it's executed in the output directory.
-
Use the -t argument to specify a local directory for temp files. Default is to use the /tmp/ folder, which might have low space on cluster (unless tmpspace is specified in cluster_config.yaml)
-
Manual interruption of the workflow: Simple Ctrl+C is enough to cancel/inturrupt the workflow. However, in some cases re-running the workflow after inturruption might fail with message "Locked working directory". In that case, please run the workflow with
-s ' --unlock'once.
Major outputs of the workflow are:
- Transcript compatibility counts (TCC) in folder
<outdir>/transcripts_quant/<sample>/eq_counts/tcc.mtx - Gene counts in folder
<outdir>/transcripts_quant/<sample>/gene_counts/gene.mtx - RNA velocity output in folder
<outdir>/velocity_output(normal/filtered loom files, velocity plots)