Skip to content

xindong95/DEAP

BranchesTags

Repository files navigation

DEAP (Alpha)

Introducton

Differential Expression Analysis Pipeline(DEAP) is a pipeline that can process expression data that RNA-seq or MicroArray experiments generated. The pipeline is built using snakemake, which allows for ease of use, optimal speed, and highly modular code that can be further added onto and customized by experienced users.

Basic Usage

We suppose you start your project with PROJECT folder:

mkdir PROJECT
cd PROJECT
git clone [email protected]:xindong95/DEAP.git
conda env create -f DEAP/environment.yaml # this might take long time based on your network environment, use mamba is recommended
conda activate DEAP

The data can be stored anywhere with the given path in the config.yaml file, but the best practice is putting all your data in the data folder.

mkdir data

You can download reference files here. All compressed files should be extracted to the ref_files folder.

mkdir ref_files
cd ref_files
wget https://XXXXXXX/hg38.tar.gz
wget https://XXXXXXX/mm10.tar.gz
wget https://XXXXXXX/GPL.tar.gz
tar xvzf hg38.tar.gz
tar xvzf mm10.tar.gz
tar xvzf GPL.tar.gz
cd ..

If you have already downloaded reference files on your server or local machine, we recommend you link the reference files by the soft link:

ln -s /fs/home/dongxin/Files/DEAP_ref_files ref_files

Copy needed config files to the current folder.

cp DEAP/metasheet.csv .
cp DEAP/ref.yaml .
cp DEAP/config.yaml .

Then your folder structure would be like this:

PROJECT/
├── data (store your data)
├── ref_files
├── DEAP (this one is the git repo)
├── metasheet.csv
├── ref.yaml
└── config.yaml

Please modify your config.yaml and metasheet following the instructions in Files Format part. After modifing config.yaml and metasheet.csv, type:

snakemake -s DEAP/DEAP.snakefile

Then the pipeline will start running for analysis. If you only want to check the tasks scheduling and command used in the analysis, you can type:

snakemake -s DEAP/DEAP.snakefile -np

Other snakemake advanced instructions can be found here.

All Data Comes From GEO

Gene Expression Omnibus (GEO) database stores a huge number curated gene expression DataSets, as well as original Series and Platform records. If you want to process data all from GEO, you can use the geo_start.py to help you download data and configure the pipeline.

usage: geo_start.py [-h] [-s SAMPLECONFIG] [-m METASHEET] [-e COMMAND]
                    [-j JOBS]

optional arguments:
  -h, --help            show this help message and exit
  -s SAMPLECONFIG, --sampleconfig SAMPLECONFIG
                        The path of new config file with samples.
  -e COMMAND, --command COMMAND
                        Other command you want to submit to snakemake,
                        should be quoted. This would overwrite "-j"/
                        "--jobs". Eg. "-npr"
  -j JOBS, --jobs JOBS  The cores you want to provide to snakemake

The basic configuration is the same as mentioned in Basic Usage part.

mkdir PROJECT
cd PROJECT
git clone [email protected]:xindong95/DEAP.git
conda env create -n DEAP -f DEAP/environment.yaml # this might took long time based on your network environment
conda activate DEAP

# use this
mkdir ref_files
cd ref_files
wget https://XXXXXXX/hg38.tar.gz
wget https://XXXXXXX/mm10.tar.gz
wget https://XXXXXXX/GPL.tar.gz
tar xvzf hg38.tar.gz
tar xvzf mm10.tar.gz
tar xvzf GPL.tar.gz
cd ..
# or this
ln -s path/to/previous/ref_files/ ref_files

cp DEAP/ref.yaml .
cp DEAP/metasheet.csv .
cp DEAP/config.yaml .

If you have already downloaded repository and reference files, soft link would be useful for saving space:

mkdir PROJECT
cd PROJECT
ln -s /path/to/DEAP DEAP
ln -s /path/to/ref_files ref_files

In this case, you still need to modify the metasheet table to define samples relationship as instructed from Files Format first. Please noticed that all sample ID (2nd columns) should be exact accession with upper case "GSM", eg. GSM123456.
But in config.yaml file, you can skip the samples part just leave them as blank.

After the configuration of these two files, you can start download and processing by:

python DEAP/geo_start.py

This script also supports passing parameter to snakemake:

python DEAP/geo_start.py -j 8 # use 8 core to run snakemake
python DEAP/geo_start.py -e "-np" # dry-run for snakemake

You can specify the output path of new config files as well:

python DEAP/geo_start.py -s new_config.yaml

Files Format

You can take the config.yaml, metasheet.csv and ref.yaml provided under the DEAP folder as a reference.

  • config.yaml:
    Here are several keys in this yaml file.

    • metasheet: a path for a table file that defined the relationship among samples.
    • ref: a yaml file that stored the path where reference files are.
    • aligner: the align tool that you want to use for RNA-seq samples, options: "salmon" or "STAR" (Not support yet).
    • assembly: the assembly you want to use for your data, options: "mm10" or "hg38".
    • trim: DEAP use trim_galore to cut off the adapter at the distal of reads and do quality control, options: True or False.
    • batch: Whether remove batch effect by inmoose.pycombat, options: True or False.
    • lisa: LISA is used to determine the transcription factors and chromatin regulators that are directly responsible for the perturbation of a differentially expressed gene set. option:True or False.
    • samples: Can include many samples in this key. For each child key, you can write one fastq file for single-end samples or two fastq files for pair-end samples. You can remove the whole key if you use geo_start.py to run the pipeline.

  • metasheet.csv:
    This is a table that define samples' relationship and the way you would like to do differential expression genes among samples. Except compare_XXX, other headers should not be modified.

    • The 1st column (run) is the run name. Samples of a run(or batch) should share the same name, though they could be divided into several rows for different FASTQ files or CEL files.
    • The 2nd column (sample) should always be samples ID. This sample ID must precisely match the sample names configured in the config yaml file. If you use geo_start.py to run the pipeline, please make sure all sample IDs are exact accession begin with upper-case "GSM", e.g. GSM123456.
    • The 3rd column (experment_type) define the experiment type of data, please choose one of {RS, MA_A, MA_O, MA_I}. RS for RNA-seq, MA_{} for microarray data generated by Affymatrix, Oligo or Illumina (Not support yet).
    • The 4th column (platform): You are supposed to write the platform each sample used. We only use the first one platform you defined in the same run but still recommend you fulfill the whole table. For RNA-seq, the platform will be used for the reads group. For microarray data, the platform should use the GPL file you downloaded from GEO, e.g., GPL570. Please make sure the GPL you write in the table has the matched file which named by GPL{XXXX}.txt in ref_files/GPL/. The {XXXX} is the number of this platform.
    • The 5th column (condition): The annotation of each sample. It would help if you wrote some string in this column. For example, "Control", "RNAi", "Treat", "ESR1i", etc. Attention: Do NOT contain any _ or - in this column.
    • The 6th column (batch): The batch information of each sample. Same batch using same number is acceptable.
    • The 7th column (Any name): The samples that you want to perform a Differential Expression to using limma and DEseq. The "control" should be marked with a 0, and the "treatment" should be marked with a 1. Recommand using string like {Treat}_vs_{Control} for clarity. You are allowed to have as many {Treat}_vs_{Control} columns as you want at the end.

  • ref.yaml:
    Here you are allowed to define your reference files, includes GPL files, hg38 index, and mm10 index files.

Appendix

LISA installation (updating to lisa2, unfinished)

LISA isn't a necessary component of DEAP, and you needn't install it if you do not plan to run it - set lisa in config as False. Otherwise, please install LISA based on the following instruction.

Since LISA also uses snakemake to build the workflow, to avoid the version conflict, we recommend creating a new environment for LISA. Here is the GIT of LISA.

  1. Install LISA by conda:

    conda create -n lisa python=3.6
conda activate lisa
conda install -c qinqian lisa

  2. Download reference files of LISA:

    wget http://lisa.cistrome.org/cistromedb_data/lisa_v1.0_hg38.tar.gz
wget http://lisa.cistrome.org/cistromedb_data/lisa_v1.1_mm10.tar.gz

  3. Extract files and update LISA configuration: The files can not move after lisa_update_conf, and please extract them to a proper position.

    tar xvfz lisa_v1.0_hg38.tar.gz
lisa_update_conf --folder hg38/ --species hg38
# or
tar xvfz lisa_v1.0_mm10.tar.gz
lisa_update_conf --folder mm10/ --species mm10

  4. Exit LISA environment and activate DEAP environment:

    conda deactivate
conda activate DEAP

Reference building

About

Differential Expression Analysis Pipeline

Topics

snakemake pipeline biology rnaseq microarray illumina affymetrix

Resources

Readme

License

GPL-3.0 license
Activity

Stars

0 stars

Watchers

1 watching

Forks

1 fork
Report repository

Releases

No releases published

Packages

No packages published

Contributors 2

  •  
  •  

Languages