/CRISPR-tools

Pipeline for CRISPR-Cas9 screen analysis

Primary LanguagePythonMIT LicenseMIT

made-with-pythonGitHub license

CRISPR screen bioinformatic pipeline

This bioinformatic pipeline will automate analysis of NGS data from CRISPR-Cas9 screen experiments. It can use MAGeCK or BAGEL2 for statistical analysis.

Index

  1. Software dependencies
  2. Installation
  3. Configuration
  4. Usage
  5. Output

Software dependencies:

Installation:

Installation from the command line:

git clone https://github.com/niekwit/CRISPR-tools.git

The CRISPR-tools directory can be permanently added to $PATH by adding the following line to your ~/.bashrc:

export PATH=/home/path/to/CRISPR-tools:$PATH

OPTIONAL: to enable auto-completion of the command line options for the CRISPR library and analysis (-l/--library and -a/--analysis), add this line to your ~/.bashrc file:

source /path/to/CRISPR-tools/auto-complete.sh

To install all software dependencies, make sure pip3 and a JRE (Ubuntu/MacOS) are installed, and then run the setup.py file from the command line as follows:

python3 setup.py

Configuration:

CRISPR libraries can be configured in the library.yaml file (located in the CRISPR-tools folder), as follows:

bassik:
  fasta: /home/niek/Documents/references/fasta/Human/Bassik-library/bassik_lib.fasta
  index_path: /home/niek/Documents/references/bowtie2-index/bassik/bassik
  read_mod: clip
  clip_seq: GTTTAAGAGCTAAGCTGGAAACAGCATAGCAA
  sg_length:
  species: human
moffat_tko1:
  fasta:
  index_path: "/home/niek/Documents/references/bowtie2-index/moffat_tko1/moffat_tko1"
  read_mod: trim
  clip_seq:
  sg_length: 20
  species: human

Explanation of library.yaml:

  • The path to the fasta file must be set with fasta (fasta files for a variety of CRISPR libraries can be found in the Addgene_CRISPR_libraries_FASTA folder).
  • The entry for index_path can be left empty, as the Bowtie2 index can be made during the analysis (will be added to this yaml file automatically)
  • If a CRISPR library has a fixed sgRNA length, then the length of the sgRNA must be set with the sg_length variable. Additionaly, set read_mod as "trim".
  • If a CRISPR library has variable sgRNA lengths, then read_mod should be set and "clip" and clip_seq should contain the vector sequence downstream of the sgRNA sequence.

Note: before the first run with any CRISPR library only the fasta file has to be provided, as the index file will be created if it is missing, and will be added to the library.yaml file.

Important: when a variable is not used (e.g. clip_seq for a fixed sgRNA length CRISPR library), it should be left empty, see example.

Usage:

  1. Create a main folder (can be any name) for the analysis that contains the subfolder raw-data, which contains the fastq.gz files.

  2. If you want to rename your sequencing files (the files names will be used as sample names for the MAGeCK analsysis so it is recommended to abbreviate them), then this can be set with the rename.config file that should be located in the main analysis folder. On each line put the existing file name and the desired new file name, separated by a semi-colon (do not include any white space), for example:

S25_S2_L001_R1_001.fastq.gz;S25.fastq.gz
L8_S1_L001_R1_001.fastq.gz;L8.fastq.gz
S8_S3_L001_R1_001.fastq.gz;S8.fastq.gz
S15_S4_L001_R1_001.fastq.gz;S15.fastq.gz

The rename.config file should be placed in the main analysis folder.

If your samples contain sequencing data from amplifications of the CRISPR library itself, then these can be named pre and post, with pre being the pre-amplification DNA (i.e. what was delivered from Addgene), and with post being the post-amplification DNA (i.e. your own maxiprep of the CRISPR library). Renaming these samples in this way, will trigger a comparative analysis of these two samples that will show any skew in sgRNA numbers (depicted by the GINI index). A good library amplification will maintain the same sgRNA number skew as the original prep.

  1. If you want to perform a comparative analysis between samples using MAGeCK or BAGEL2, then a stats.config file has to be created, for example:
t;c
S8;L8
S15;L8
S25;L8

c: reference sample, t: test sample. In the MAGeCK output file: neg rank(genes that drop out in test sample)/pos rank(genes that are overrepresented in test sample).

The stats.config file should be placed in the main analysis folder.

  1. To get an overview of all the options for the CRISPR analysis, type path/to/crispr.py -h, --help in the command line:
usage: crispr.py [-h] -l {CRISPR library}
                 [-t <int>] [-r] [-m N] [-a {mageck,bagel2}] [-f <FDR value>]
                 [-c <CCLE cell line>] [--go] [--skipfastqc]

optional arguments:
  -h, --help            show this help message and exit
  -l {CRISPR library}, --library {CRISPR library}
                        CRISPR library
  -t <int>, --threads <int>
                        Number of CPU threads to use (default is 1). Use max
                        to apply all available CPU threads
  -r, --rename          Rename fastq files according to rename.config
  -m N, --mismatch N    Number of mismatches (0 or 1) allowed during alignment
  -a {mageck,bagel2}, --analysis {mageck,bagel2}
                        Statistical analysis with MAGeCK or BAGEL2. Default is
                        MAGeCK
  -f <FDR value>, --fdr <FDR value>
                        Set FDR cut off for MAGeCK hits (default is 0.25)
  -c <CCLE cell line>, --cnv <CCLE cell line>
                        Activate CNV correction for MAGeCK/BAGEL2 with given
                        cell line
  --go                  Gene set enrichment analysis with enrichR
  --skip-fastqc          Skip FastQC/MultiQC analysis

To start an analysis, for example with the Bassik whole-genome CRISPR library, navigate to main analysis folder in the command line and run:

crispr.py -l bassik -r -t max

This initiates a run that will rename your samples according to rename.config, allows no mismatches during alignment, will use all available CPU threads for the analysis, and uses MAGeCK for statistical analysis. The FDR cut off to determine significant hits is set at 0.25, and can be changed with the -f/--fdr flag. If you also want to use BAGEL2 for statistical analysis afterwards, simply run:

crispr.py -l bassik -a bagel2

This will only run BAGEL2 and skip all steps that are common with MAGeCK.

Both MAGeCK and BAGEL2 have the possibility to correct the effects from copy number variations. To enable this feature the -c/--cnv flag can be added to the command line.

Finally, if you decide to add more samples to the analysis folder, or more comparisons in the stats.config file, then these can be analysed without overwriting previous data.

Output:

Several folder/files will be generated:

  • fastqc: contains FastQC and MultiQC analyses on the fastq.gz files.
  • count: contains the sgRNA counts in individual files, and counts of all files collated in one file (counts-aggregated.tsv, MAGeCK input file). It also contains a normalised version of counts-aggregated.tsv.
  • library analysis: contains the analyses of the pre and post library amplification samples.
  • mageck: contains the MAGeCK output files. It will also contain plots of the results with the top 10 genes marked.
  • bagel2: contains the BAGEL2 output files. It will also contain plots of the results with the top 10 genes marked.

Output examples:

Example 1: analysis of Yusa Mouse CRISPR library amplification

If your experiment consists of just pre and post library amplification samples (e.g. the Yusa Mouse library), rename your fastq files to pre.fq.gz and post.fq.gz, and then start the analysis as follows:

crispr.py -l yusa-mouse

This will run FastQC/MultiQC to check the quality of the fastq files, and because there are no experimental samples, only the CRISPR library analysis will be run. The count directory will contain the alignment-rate.pdf file that gives an overview of the alignment rates of all the fastq files:

The library-analysis directory will contain four files:

  1. lorenz-curve.pdf: this will show the Lorenz curve and the Gini indices for each sample