Across cell type NGS data normalization¶

usage: S3norm.py [-h] [-j JID] -f INPUT [--smoothLength SMOOTHLENGTH]
                 [--binSize BINSIZE] [--peak_dir PEAK_DIR]
                 [--file_name_pattern FILE_NAME_PATTERN] (-bam | -bw | -bdg)
                 [-g GENOME] [-s CHROM_SIZE] [-b BLACK_LIST]

NGS read counts normalization

optional arguments:
  -h, --help            show this help message and exit
  -j JID, --jid JID     enter a job ID, which is used to make a new directory.
                        Every output will be moved into this folder. (default:
                        S3norm_yli11_2022-05-23)
  -f INPUT, --input INPUT
                        TSV file, 1 or 2 columns. If the 2nd col is available,
                        it will be used as control. (default: None)
  --smoothLength SMOOTHLENGTH
  --binSize BINSIZE     bin size for s3norm (default: 20)
  --peak_dir PEAK_DIR
  --file_name_pattern FILE_NAME_PATTERN
  -bam                  input files are bam, bam need index (default: False)
  -bw                   input files are bigwig (default: False)
  -bdg                  input files are bedgraph (default: False)

Genome Info:
  -g GENOME, --genome GENOME
                        genome version: hg19, hg38, mm9, mm10. By default,
                        specifying a genome version will automatically update
                        index file, black list, chrom size and
                        effectiveGenomeSize, unless a user explicitly sets
                        those options. (default: hg19)
  -s CHROM_SIZE, --chrom_size CHROM_SIZE
                        chrome size (default: /home/yli11/Data/Human/hg19/anno
                        tations/hg19.chrom.sizes.sorted)
  -b BLACK_LIST, --black_list BLACK_LIST
                        Blacklist file (default: /home/yli11/Data/Human/hg19/a
                        nnotations/hg19.blacklist.bed)

Summary¶

S3norm can perform read counts normalization across different cell types. It uses a monotonic nonlinear data transformation method to match signals in both the peak regions and the background regions differently, such that both sequencing depth and signal-to-noise ratio between data sets can be simulatenously normalized.

Ref:

https://github.com/guanjue/S3norm

https://github.com/YichaoOU/S3norm

Input¶

A 1 column or 2 columns tsv file containing your signal data with or without control data.

For example

chip_seq_1      control_1
chip_seq_2      control_2
chip_seq_3      control_1
atac_seq_1

Input files can be one of the 3 types, but not mixed types:

Bam (with .bai index file in the same dir)
bigwig
bedgraph

Usage¶

Put all your input data in the same working directory and inside this dir, do the following:

hpcf_interactive

module load python/2.7.13

S3norm.py -f input.list -bam

To make S3norm run faster, we can only normalize signals around the peak region:

S3norm.py -f v15input.list --peak_dir ../cut_run_rfeng_2021-05-20/peak_files -bam

Output¶

The output files are S3norm-normalized bigwig files, located at $jid/S3norm_NBP_bedgraph/*.bw

Common problems¶

1. no outputs.¶

This could be caused by out of memory error if you have many files to be normalized at the same time. It also depends on number of entries (i.e., resolution, binsize) in your bedgraph file. In one example I have, 6 files with hg19 binsize=100bp, took almost 50G memory to run and finish in about 3 hours.

You shouldn’t have this problem if your data looks similar to the example above because by default, we require 80G memory.

2. input bw or bdg¶

--bdg not implemented. If you inputs are bigwig, use S3norm.py -f input.list -bw, similarly, if inputs are bedgraph, use -bdg

Comments¶

code @ github.