sctools

Tools for single-cell RNA-seq analysis

🚨 This package is no longer maintained. Some functionality has been moved to sinto package. 🚨

Installation

git clone https://github.com/timoast/sctools.git
cd sctools
python -m pip install .

Features

Count SNPs

If your data contains a mixture of two genotypes, you can count UMIs derived from each genotype in each cell using the sctools countsnps command.

$ sctools countsnps -h
usage: sctools countsnps [-h] -b BAM -s SNP -o OUTPUT [-c CELLS] [-p NPROC]

Count reference and alternate SNPs per cell in single-cell RNA data

optional arguments:
  -h, --help            show this help message and exit
  -b BAM, --bam BAM     Input bam file (must be indexed)
  -s SNP, --snp SNP     File with SNPs. Needs chromosome, position, reference,
                        alternate as first four columns
  -o OUTPUT, --output OUTPUT
                        Name for output text file
  -c CELLS, --cells CELLS
                        File containing cell barcodes
  -p NPROC, --nproc NPROC
                        Number of processors (default = 1)

Genotype cells

If your data is a mix of two genotypes, you can assign genotype labels to cells based SNP counts calculated using sctools countsnps, using the sctools genotype command.

This runs an iterative density-based clustering of cells based on SNP UMIs using DBSCAN.

$ sctools genotype -h
usage: sctools genotype [-h] -s INFILE -o OUTFILE [-n NPROC] [-p] [-d]
                        [--eps_background EPS_BACKGROUND]
                        [--eps_cells EPS_CELLS]
                        [--min_samples_background MIN_SAMPLES_BACKGROUND]
                        [--min_samples_cells MIN_SAMPLES_CELLS]

Genotype cells based on SNP UMI counts.

optional arguments:
  -h, --help            show this help message and exit
  -s INFILE, --infile INFILE
                        SNP UMI counts for each genotype
  -o OUTFILE, --outfile OUTFILE
                        Name for output text file
  -n NPROC, --nproc NPROC
                        Number of processors (default = 1)
  -p, --plot            Plot results
  -d, --downsample      Do not downsample cells before detecting background
                        cluster
  --eps_background EPS_BACKGROUND
                        DBSCAN epsilon value for background cell clustering
  --eps_cells EPS_CELLS
                        DBSCAN epsilon value for background cell clustering
  --min_samples_background MIN_SAMPLES_BACKGROUND
                        Minimum number of cells in each cluster for background
                        cell clustering
  --min_samples_cells MIN_SAMPLES_CELLS
                        Minimum number of cells in each cluster for background
                        cell clustering

Filter barcodes

If you want to extract reads from a subset of cell barcodes, you can do so using the sctools filterbarcodes command.

$ sctools filterbarcodes -h
usage: sctools filterbarcodes [-h] -b BAM -c CELLS -o OUTPUT [-s]
                                 [-p NPROC]

Filter reads based on input list of cell barcodes

optional arguments:
  -h, --help            show this help message and exit
  -b BAM, --bam BAM     Input bam file (must be indexed)
  -c CELLS, --cells CELLS
                        File containing cell barcodes. Can be gzip compressed
  -o OUTPUT, --output OUTPUT
                        Name for output text file
  -s, --sam             Output sam format (default bam output)
  -p NPROC, --nproc NPROC
                        Number of processors (default = 1)

Count edited transcripts

Edited RNA transcripts can be counted per editing position per cell using the sctools countedited command.

$ sctools countedited -h
usage: sctools countedited [-h] -b BAM -e EDIT -o OUTPUT [-c CELLS]
                              [-p NPROC]

Count edited transcripts per gene per cell in single-cell RNA data. Output is
a matrix of positions by cells.

optional arguments:
  -h, --help            show this help message and exit
  -b BAM, --bam BAM     Input bam file (must be indexed)
  -e EDIT, --edit EDIT  File with edited base coordinates. Needs chromosome,
                        position, reference, alternate as first four columns
  -o OUTPUT, --output OUTPUT
                        Name for output text file
  -c CELLS, --cells CELLS
                        File containing cell barcodes to count edited bases
                        for. Can be gzip compressed (optional)
  -p NPROC, --nproc NPROC
                        Number of processors (default = 1)

Replace SNP bases in reference genome

Nucleotides in the reference genome can be replaced at SNP positions with the IUPAC code encoding both/all possible nucleotides using the sctools replace command. The reason for doing this is to decrease any reference-bias in aligning cDNA reads.

$ sctools replace -h 
usage: sctools replace [-h] -g GENOME -s SNP -o OUTPUT

substitute SNPs into reference genome

optional arguments:
  -h, --help            show this help message and exit
  -g GENOME, --genome GENOME
                        reference genome fasta file
  -s SNP, --snp SNP     snp file in tsv format
  -o OUTPUT, --output OUTPUT
                        output filename

Import sctools as a python module

You can also use the sctools functions interactively or in python scripts by importing sctools as a module.

For example, to determine cell genotypes more interactively (rather than running sctools genotype on the command line), you can do something like this:

from sctools import genotype
import pandas as pd

data = pd.read_csv("snps.tsv")

geno = genotype.Genotype(data)                       # initiate the genotype class
geno.transform_snps()                                # log transform SNP UMI counts
geno.filter_low_count(min_log10_count=1)             # remove cells with less than log10(UMI) = 1
geno.detect_total_background(eps=1, min_samples=500) # detect background cell cluster
geno.segment_cells()                                 # downsample cells to equal numbers for each genotype
geno.detect_cells(eps=0.3, min_samples=100)          # detect cell clusters
geno.label_barcodes()                                # attach genotype labels to cell barcodes

You can also use functions in sctools to read and subset the gene expression matrix output by CellRanger in sparse matrix format.

from sctools import sctools

tenx = sctools.SC()
tenx.read_10x("filtered_gene_bc_matrices/genome")
subset_cells = tenx.subset(cells = ['AAGGAGCCAGCTATTG', 'AAACCTGGTAGCTAAA'])
subset_cells.counts.toarray()