envest/single-cell_lineage

Lineage tracking from single-cell DNA mutations

single-cell_lineage's Issues

Genotype cells with set of SNPs and un-pool samples

A set 22 SNPs (subset: 7 SNPs) were included in a the Mission Bio panel for VHR-ALL data.

These SNPs can be used to differentiate between pooled samples.

Each patient has one non-pooled sample (usuallly diagnosis) and some samples were pooled to create enough material for sequencing (MRD stage).

1. Genotype every cell in each non-pooled sample at all SNPs
2. Explore consistency of genotypes across cells
3. Derive a consensus genotype for every non-pooled sample
4. Genotype every cell in each pooled sample at all SNPs
5. Match each cell's genotype to a diagnosis sample

Ultimately, we need a table like this:

sample	stage	pool	cell
sample_1	diagnosis	NA or pool_n	cell_barcode_1
...

Some cells may not match perfectly -- we can provide a best-guess and a confidence level. We may exclude cells with confidence below a threshold.

It's possible we may want to un-pool VCFs, but that may not be necessary.

Collect VHR-ALL mission bio metadata

We need to collect metadata for each sample, especially

• Link each sample to its VCF file
• Split up pooled samples
• Split up column describing mixture of tumor and normal cells

Create and document reproducible environment

My initial expectation is this environment should include:

• samtools bcftools
• Python: pysam pyvcf
• R: tidyverse

Ultimately working with a nextflow singularity container will translate to HPC

Create script to download, sort, and index mission bio VCF files

Mission Bio VCFs can be downloaded using wget, which uses a URL and needs an output file name/location.

1. Download using command line arguments URL and file_name
2. Use bcftools sort to create a sorted bcf.gz file
3. Use bcftools index to index the sorted `bcf.gz. file in the same location

Select variants from mission bio BCF

Mission Bio data includes variants not included in the target list due to off-target coverage.

This script should select --targets targets.tsv from a --input_file sorted.bcf.gz and storing in --output_file target_description.bcf.gz.

bcftools view takes a BCF and target list as input and outputs just our variants of interest.

Download bam file /index from url

We need to download bam files from remote sites. Sometimes they will have an index (also download that) but sometimes not (create index with samtools index).

Tapestri scDNA QC files

For each tapestri scDNA VCF file, we want to include our own QC report. It can include metrics already provided by the pipeline, but expand to focus on metrics for each variant and barcode, including proportion of genotypes called in either dimension.

Create two R markdown notebooks:

• Sample level QC
• Project level QC

Inputs:

1. File naming sheet
2. Patient info
3. Variant list

Notebooks can be shared with collaborators as something like

• patient.sample.scdna_qc_report.html
• project_name.scdna_qc_overview.html