Phen2Gene is a phenotype-driven gene prioritization tool, that takes HPO (Human Phenotype Ontology) IDs as inputs, searches and prioritizes candidate causal disease genes. It is distributed under the MIT License by Wang Genomics Lab. Additionally we have provided a web server and an associated RESTful API service for running Phen2Gene. Finally, a mobile app for Phen2Gene and several other genetic diagnostic tools from our lab is being tested and will be available soon.
If you do not wish to use Anaconda, simply install the packages in the file environment.yml using pip. If you use conda, some packages may not properly install without updating conda using conda update conda first.
If you are lucky enough to have Docker or some equivalent like Singularity or Podman installation is easy as pie, just download the docker image with the following command:
docker pull genomicslab/phen2gene
Test out your Docker image with the below commands:
On Unix/Linux:
docker run -it --rm -v $PWD/out:/code/out -t genomicslab/phen2gene -m HP:0001250 -out out/prioritizedgenelist
On Windows Powershell:
docker run -it --rm -v ${PWD}/out:/code/out -t genomicslab/phen2gene -m HP:0001250 -out out/prioritizedgenelist
As of the Jan 2021 version of the HPO2Gene KnowledgeBase, if you see in out/prioritizedgenelist/output_file.associated_gene_list:
Rank Gene ID Score Status
1 KCNQ2 3785 1.0 SeedGene
2 KCNQ3 3786 0.936339 SeedGene
3 UBE3A 7337 0.93565 SeedGene
4 MECP2 4204 0.89883 SeedGene
5 FGFR2 2263 0.830351 SeedGene
You have succeeded. The arguments described below in this document will work if you replace python3 phen2gene.py with docker run -it --rm -v $PWD/out:/code/out -t phen2gene.
First, install Miniconda, a minimal installation of Anaconda, which is much smaller and has a faster installation. Note that this version is meant for Linux below, macOS and Windows have a different script:
curl -O https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
bash Miniconda3-latest-Linux-x86_64.sh
Go through all the prompts (installation in $HOME is recommended).
After Anaconda is installed successfully, simply run:
git clone https://github.com/WGLab/Phen2Gene.git
cd Phen2Gene
conda env create -f environment.yml
conda activate phen2gene
bash setup.sh
This software can be used in one of three scenarios:
- Ideally, you have a list of physician-curated HPO terms describing a patient phenotype and a list of potential candidate disease variants that overlap gene space and you want to narrow down the list of variants by prioritizing candidate disease genes, often in tandem with variant prioritization software, which cannot as of yet score STR expansions or SVs unlike Phen2Gene which is variant agnostic.
- You do not have variants, but you have HPO terms and would like to get some candidate genes for your disease that you may want to target sequence, as it is much cheaper than whole-genome or whole-exome sequencing.
- If you have clinical notes, you can use tools like EHR-Phenolyzer or Doc2HPO for processing clinical notes into HPO terms using natural language processing (NLP) techniques, then apply scenario 1 or 2 as relevant.
Input files to Phen2Gene should contain HPO IDs, separated by UNIX-recognized new line characters (i.e., \n).
Alternatively you can use a space separated list of HPO IDs on the command line.
- Input HPO IDs via input file (typical use case)
python3 phen2gene.py -f example/HPO_sample.txt -out out/prioritizedgenelist
- Input HPO IDs via input file, and candidate gene list file (another common use case)
python3 phen2gene.py -f example/HPO_sample.txt -out out/prioritizedgenelist -l example/1000genetest.txt
- Use Skewness and Information Content
-w skuses a skewness-based weighting of genes for each HPO term (default, and recommended)-w wand-w icdo not use skew, but utilize information content in the tree structure (slightly worse performance)-w uis unweighted
python3 phen2gene.py -f example/HPO_sample.txt -w sk -out out/prioritizedgenelist
- Run Phen2Gene with verbose messages
python3 phen2gene.py -f example/HPO_sample.txt -v -out out/prioritizedgenelist
- Input HPO IDs manually, if desired
python3 phen2gene.py -m HP:0000021 HP:0000027 HP:0030905 HP:0010628 -out out/prioritizedgenelist
- Add H2GKB location manually, if desired
python3 phen2gene.py -f example/HPO_sample.txt -d full_path_to_H2GKB.zip_extraction_folder -out out/prioritizedgenelist
Examples of how to use the Web Server and the RESTful API can be found in the Docs.
Please use the Phen2Gene issues page if you have any questions!
In order, run:
bash setup.sh # You can skip it if you ran it in the installation.
bash runtest.sh
If you only want the benchmark data and nothing else:
bash getbenchmark.sh /directory/to/download/to
The figures are in the folder figures.
After changing the code example/ANKRD11example.sh so the ANNOVAR db is built where you would like it, simply run:
bash example/ANKRD11example.sh
Going through the code in example/ANKRD11example.sh, first one downloads a list of candidate variants from the article referenced in the manuscript where the patient has KBG syndrome.
Then, we annotate with ANNOVAR to retrieve gene annotations for these variants, functional consequence information (exonic, intronic, nonsynonymous), amino acid change information, and population frequency.
We next filter out common variants (>1% in gnomAD 2.1.1) and use Phen2Gene to rank the candidate genes based on HPO terms.
Combining this information with the variants, we can re-rank Phen2Gene's candidate list as in the script filterbyannovar.py and discover that the variant for the causal gene ANKRD11 is now ranked number 1 after being ranked number 2 by HPO term. The number 1 ranked gene by HPO, VPS13B, is filtered out because the only candidate variant (8-100133706-T-G) has an extremely high allele frequency in gnomAD(74%!).
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent