Metadensity

Visualize eCLIP/STAMP metagene density to predict RNA binding protein function [Read the Doc] https://metadensity.readthedocs.io/en/latest/

Install

The environment is available at DockerHub

git clone https://github.com/algaebrown/Metadensity.git
cd Metadensity
# build your own environment!
conda env create -n Metadensity --file environment.yaml
conda activate Metadensity
# copy genome coordinate
cd Metadensity
pip install -e .

Step 1: download data and setup paths to annotations in config/

Metadensity requires several annotations to work. You need to point those files in config/*.ini. see config/hg38.ini as an example. These information are genome coordinate, species dependent. So you can keep each species with a seperate .ini.

	Description	Link to download	Essential to run
GENOME_FA	fasta for the entire genome sequence	https://www.ncbi.nlm.nih.gov/genome/guide/human/	YES
GENCODE	gff3 annotation of exon, intron, gene, transcripts etc	https://www.gencodegenes.org/human/	YES
BRANCHPOINT	branchpoint annotation	https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4315302/	NO
BRANCHPOINT_PRED	machine learning predicted branchpoints	https://pubmed.ncbi.nlm.nih.gov/29092009/	NO
POLYA	annotation of polyA sites and signals from polyASite	https://polyasite.unibas.ch/atlas	NO
MIRNA	annotation of microRNA	https://www.mirbase.org/	NO
SNORNA	annotation of snoRNA	http://scottgroup.med.usherbrooke.ca/snoDB/	NO
LNCRNA	annotation of lncRNA	https://lncipedia.org/	NO
TRANSCRIPT	gff3 annotation from GENCODE, containing only "transcript"	use this script to generate from GENCODE and UCSC canonical transcript list or download them *nonpickle here	YES
FEATURE	gff3 annotation from GENCODE, containing only "exon", "CDS", "UTR" and created "introns"	use this script to generate from GENCODE and UCSC canonical transcript list; or download them *nonpickle	YES
DATADIR	parsed information of above, python .pickle file	https://www.dropbox.com/sh/hoya37n9pmuqd4l/AABBSpcpjFYIUFWMdIRuJtU4a?dl=0

Advanced usage:

if you are using some coordinate that we don't have precomputed "DATADIR", please check out this notebook on how to build one yourself.

Step 2: Command line usage: the most vanilla functions

python scripts/run_metadensity_vanilla.py -h

Options:
  -h, --help            show this help message and exit
  -i CSV, --csv=CSV     .csv file containing all CLIP files
  -u UID, --uid=UID     unique ID(uid) to the CLIP in the csv you want to run
                        on
  -t TRANSCRIPT_LIST, --transcript_list=TRANSCRIPT_LIST
                        list of transcript ids to include in the calculation,
                        if not specify, use peak-containing transcripts
  -o OUTDIR, --out=OUTDIR
                        output path (figures and arrays)
  -s, --single end      Whether your CLIP is single end. Affects Metatruncate
                        objects
  -c CONFIG, --config=CONFIG
                        file path to the config file, genome coordinate
                        specific
  --stat=STAT           choose [mean,median]
  --background_method=BG
                        how you want to compute IP to INPUT, choose [relative
                        information,substraction,None]
  --normalization       whether to average the signal in a transcript
  --truncation          Use truncation instead of the entire read

This will run the most vanilla functions of Metadensity

Here we provide some test data to run the script.

cd Metadensity
# download the data
wget https://www.dropbox.com/s/cgkeuqr0cjif558/test_data.tar.gz

# uncompress
tar -xvzf test_data.tar.gz

# modify paths is menifest.csv to correspond to your directory
nano test_data/menifest.csv

# run
cd scripts/
python run_metadensity_vanilla.py -i ../test_data/menifest.csv -u SF3B4_test -o ../test_data --config=../config/hg38.ini

Step 3: Advanced usage in jupyter notebook

checkout example jupyter notebooks