=# Repository of code used to do sortChIC processing and analysis

This is a repository of all scripts used in the sortChIC paper. For a curated set of analysis vignettes and example pipelines, visit https://github.com/jakeyeung/sortchicAnalysis

processing_scripts:

• These are scripts used for processing data, mainly through through the high performance computing cluster

rstudio_scripts:

• These are interactive scripts that loads data onto memory and performs downstream analyses on an Rstudio server

Installing dependencies

• Some packages are install.packages() some are through bioconductor, except JFuncs is installed through github devtools::install_github("jakeyeung/JFuncs")

Example of running pipeline:

In example_processing_pipeline/ there is a full example that takes fastq files, demultiplexes, trims, maps, tags bam files, generates count tables, and runs count tables through dimensionality reduction.

Example scripts are ordered from 1 to 6, which are run in ascending order. SingleCellMultiOmics scripts used v0.1.25 from https://github.com/BuysDB/SingleCellMultiOmics/releases/tag/v0.1.25 (SCMO)

(1) demultiplex usingdemux.py (SCMO)

1-run_demux.sh: runs demux.py from SCMO. Adapters, molecule barcodes, and cell barcodes are removed from reads and encoded into the fastq headers.

(2) remove adapters using cutadapt

2-trim_fastq.sh: runs cutadapt to remove Illumina adapters

(3) map trimmed fastqs using bwa

3-map-fastq.sh: uses bwa to map trimmed fastqs

(4) tag mapped bam files using bamtagmultiome.py (SCMO)

4-sort_index_tag_bam.sh: uses SCMO to read fastq headers and mapping information to record PCR duplicates, cell barcodes, cut locations, and other meta information. See https://github.com/BuysDB/SingleCellMultiOmics/blob/master/TAGS.MD for a full description of different tags and their descriptions.

(5) generate count tables using bamToCountTable.py

5-make_count_tables*.sh: reads tagged bams and outputs counts falling into genomic regions. Uses --filterXA to ignore alternative hits, --minMQ for read quality, --dedup to remove PCR duplicates, --r1only to count fragments rather than reads (i.e. there are often two reads per fragment), -blacklist to ignore reads falling in bad regions.

(6) run dimensionality reduction using run_LDA_model.R

6-run_LDA*.sh: runs latent Dirichlet allocation on cleaned count matrix (i.e. bad cells removed, bad regions removed).

Vignettes for exploring data

Check out "https://github.com/jakeyeung/sortchicAnalysis" for analysis vignettes

FAQ (Frequenty Asked Questions)

Where do I find the sortChIC bam files and how do I look at them at single-cell resolution?

For genome-wide analysis, it is better to look at bam files rather than count matrices (which are better for dimensionality reduction and fitting statistical models).

These bams have been already split and merged by cell types.

H3K4me3: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSM5018603

H3K27me3: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSM5018601

H3K4me1: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSM5018604

H3K9me3: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSM5018607

To get cell barcode from bam, look at the bam tag "CB", details for more tags in the github: https://github.com/BuysDB/SingleCellMultiOmics/blob/master/TAGS.MD

To extract the cell barcode from a read from bam, use pysam e.g. read.get_tag("CB")

Avoid the qChIC bam files, they are old.