cellSNP - Analysis of expressed alleles in single cells
Project description
cellSNP aims to pileup the expressed alleles in single-cell or bulk RNA-seq data, which can be directly used for donor deconvolution in multiplexed single-cell RNA-seq data, particularly with vireo, which assigns cells to donors and detects doublets, even without genotyping reference.
cellSNP heavily depends on pysam, a Python interface for samtools and bcftools. This program should give very similar results as samtools/bcftools mpileup. Also, there are two major differences comparing to bcftools mpileup:
cellSNP can pileup either the whole genome or a list of positions, with directly splitting into a list of cell barcodes, e.g., for 10x genome. With bcftools, you may need to manipulate the RG tag in the bam file if you want to divide reads into cell barcode groups.
cellSNP uses simple filtering for outputting SNPs, i.e., total UMIs or counts and minor alleles fractions. The idea here is to keep most information of SNPs and the downstream statistical model can take the full use of it.
News
All release notes can be found in doc/release.rst.
For computational efficiency, we initialised comments on this: doc/speed.rst
Installation
cellSNP is available through pypi. To install, type the following command line, and add -U for upgrading:
pip install cellSNP
Alternatively, you can download or clone this repository and type python setup.py install to install. In either case, add --user if you don’t have the permission as a root or for your Python environment.
From v0.1.0, cellSNP requires pysam>=0.15.2, so make sure you are using the right version of pysam. Try pip uninstall pysam and then reinstall pip install -U pysam
Quick usage
Once installed, check all arguments by type cellSNP -h (see a snapshot) There are three modes of cellSNP:
Mode 1: pileup a list of SNPs for a single BAM/SAM file
Use both -R and -b.
Require: a single BAM/SAM file, e.g., from cellranger, a list of cell barcodes, a VCF file for common SNPs. This mode is recommended comparing to mode 2, if a list of common SNP is known, e.g., human (see Candidate SNPs below)
cellSNP -s $BAM -b $BARCODE -O $OUT_DIR -R $REGION_VCF -p 20 --minMAF 0.1 --minCOUNT 20
As shown in the above command line, we recommend filtering SNPs with <20UMIs or <10% minor alleles for downstream donor deconvolution, by adding --minMAF 0.1 --minCOUNT 20
Mode 2: pileup whole chromosome(s) for a single BAM/SAM file
Don’t use -R but flexible on -b.
This mode requires inputting a single bam file with either cell barcoded (add -b) or a bulk sample:
# 10x sample with cell barcodes
cellSNP -s $BAM -b $BARCODE -O $OUT_DIR -p 22 --minMAF 0.1 --minCOUNT 100
# a bulk sample without cell barcodes and UMI tag
cellSNP -s $bulkBAM -O $OUT_DIR -p 22 --minMAF 0.1 --minCOUNT 100 --UMItag None
Add –chrom if you only want to genotype specific chromosomes, e.g., 1,2, or chrMT.
Recommend filtering SNPs with <100UMIs or <10% minor alleles for saving space and speed up inference when pileup whole genome: --minMAF 0.1 --minCOUNT 100
Note, this mode may output false positive SNPs, for example somatic variants or falses caussed by RNA editing. These false SNPs are probably not consistent in all cells within one individual, hence confounding the demultiplexing. Nevertheless, for species, e.g., zebrafish, without a good list of common SNPs, this strategy is still worth a good try, and it does not take much more time than mode 1.
Mode 3: pileup a list of SNPs for one or multiple BAM/SAM files
Use -R but not -b.
Require: one or multiple BAM/SAM files (bulk or smart-seq), their according sample ids (optional), and a VCF file for a list of common SNPs. BAM/SAM files can be input in comma separated way (-s) or in a list file (-S).
cellSNP -s $BAM1,$BAM2,$BAM3 -I sample_id1,sample_id2,sample_id3 -o $OUT_FILE -R $REGION_VCF -p 20 --UMItag None
cellSNP -S $BAM_list_file -I sample_list_file -o $OUT_FILE -R $REGION_VCF -p 20 --UMItag None
Set filtering thresholds according to the downstream analysis. Please add --UMItag None if you bam file does not have UMIs, e.g., smart-seq and bulk RNA-seq.
List of candidate SNPs
A quality list of candidate SNPs (ususally common SNPs) are important for mode 1 and mode 3. If a list of genotyped SNPs is available, it can be used to pile up. Alternatively, for human, common SNPs in population that have been idenetified from consortiums can also be very good candidates, e.g., gnomAD and 1000_Genome_Project. For the latter, we have compiled a list of 7.4 million common variants (AF>5%) with this bash script and stored in this folder.
In case you want to lift over SNP positions in vcf file from one genome build to another, see our LiftOver_vcf wrap function.
FAQ and releases
For troubleshooting, please have a look of FAQ.rst, and we welcome reporting any issue.
All releases are included in pypi. Notes for each release are recorded in release.rst.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.