PhenotypeXpression (PhenoX) is a proof of principle precision medicine tool for clinicians to quickly discover potential subtypes of known diseases. PhenoX rapidly aggregates gene expression data from multiple public studies, and then mines PubMed literature to develop novel disease subclassifications and expression profiles. By combining literature occurrences of subtype HPO and DOID information with GEO dataset expression profiles, each sub-classification of a disease is given a characteristic terminology fingerprint. Here we demonstrate the use of this tool for a common disease—Psoriasis—and identify two clusters with unique gene expression profiles.
Implemented in Python 3.6 and R. For simplicity, the install script setup.sh utilizes Anaconda3 for a self contained environment. Tested environments include Anaconda3 >= 4.3.1. To install:
A prebuilt docker image is available at https://hub.docker.com/r/ncbihackathons/phenotypexpression
docker pull ncbihackathons/phenotypexpression
docker -it ncbihackathons/phenotypexpression
python run_phenox.py -e A.N.Other@example.com "Psoriasis"
Alternatively, a local installation via a setup script should work with your Anaconda/Miniconda installation. Otherwise it will install Anaconda in your home directory. In either case, it will build a "phenoX" environment and install into that location.
git clone https://github.com/NCBI-Hackathons/phenotypeXpression.git
cd phenotypeXpression
sh setup.sh
After installation, the program can be run from the same folder:
source activate phenoX
python run_phenox.py -e A.N.Other@example.com "Psoriasis"
usage: run_phenox.py [-h] [-o prefix] [--version] -e EMAIL query_str
Subclassification of disease states based on the intersection of literature and expression
positional arguments:
query_str disease query term
optional arguments:
-h, --help show this help message and exit
-o prefix choose an alternate prefix for outfiles
--version show program's version number and exit
required arguments:
-e EMAIL NCBI requires an email for database queries
The primary inputs are a MeSH term and your email. The MeSH term should specify the parent condition for which you would like to derive subgroup information. The email is required to batch query NCBI databases.
The user can specify a prefix with -o in addition to the automatic MeSH term prefix at the beginning of each output file.
python dependencies
biopython>=1.72
spacy>=2.0.12
spacy-lookup>=0.0.2
tqdm>=4.23.4
numpy>=1.14.5
pandas>=0.23.3
wordcloud>=1.4.1
scikit-learn>=0.19.2
pydendroheatmap>=1.5
scipy>=1.1.0
conda dependencies
rpy2
r-ape
r-pvclust
r-circlize
matplotlib
readline
r-gplots
This should produce the following files in the phenotypeXpression/output directory:
- Dendgrogram of GDS with node P-values and cluster boxes.
- A word cloud for each cluster box.
- Heatmap of differentially expressed genes per GDS (columns).
- A pairwise distance graph of each GDS by absolute difference in gene sets.
- A newick tree file
- A cluster batch statistics text file.
- A cluster HPO/DOID term frequency text file.
Personalized medicine promises an almost limitless subdivision of clinical phenotypes. Accordingly, treatments can be tailored to individuals based on the specifics of their disease presentation (Mirnezami et al, 2012). In an effort to define and organize the rapid expansion of phenotypes and related conditions, a number of ontological systems have been developed, including the Online Mendelian Inheritance in Man (OMIM) (Hamosh et al, 2002), Orphanet (Weinreich et al, 2008), Human Phenotype Ontology (HPO) (Robinson et al, 2010), Human Disease Ontology (DOID) (Schriml et al, 2012), MedGen (Landrum et al, 2014), and others. These resources attempt to reach a consensus hierarchy for disease and a shared nomenclature. While unique identifiers are useful for resolving semantic differences, they do not provide an adequate description of the phenotype itself.
There has also been a steady increase in the amount of publicly available genetic expression data as public resources have scaled. The NCBI’s Gene Expression Omnibus (GEO) collects over 4000 gene expression datasets (Edgar et al, 2002). Several additional resources have been developed for expression data, such as the Human Cell Atlas (Regev et al, 2017),the Gene Expression Atlas (Kapushesky et al, 2010), the Cancer Genome Atlas (Weinstein et al, 2013), GTEx (Lonsdale et al, 2013), and others. These resources provide promising opportunities to aggregate and share expression data across multiple projects and for identifying novel phenotypes.
For the given MeSH term, PhenoX will attempt to generate all subclassifications of the condition based on the available expression data in the NCBI Gene Expression Omnibus. Each subclass will have two major features, a word cloud of phenotypic terms associated with the seach term via a literature search, and a gene expression profile of differentially expressed genes. The output will include several file types described in detail below:
The PubMed identifier cited by each GDS is retained. Differential genes conserved across entire clusters are expanded using Hugo Gene Nomenclature (HGNC) gene names and aliases (Povey et al, 2001). This expanded gene list is used with the original MeSH term to query the PubMed database for additional PubMed articles potentially related to each cluster. For each cluster, we consolidate the titles, abstracts, and keywords corresponding to associated PubMed articles, and identify HPO and DOID terms within the article texts. The Spacy Lookup extension library is used for named entity recognition (NER) based on HPO and DOID dictionary terms. Term frequency data is gathered for each article, and the counts are merged for each cluster. Word clouds are generated and used to visualize the term frequency differences between clusters, with a separate term frequency list exported as a text file.
GEO profiles matching the MeSH term are filtered using the “up down genes” filter. Resulting matches are parsed to return a list of differentially expressed gene names and corresponding GEO DataSets (GDS) (Barrett et al, 2013). All GDS are retrieved and hierarchically clustered using R-pvclust hclust (Suzuki et al, 2006); the presence or absence of each gene in a GDS is represented as a binary variable and used for clustering. GDS with only a single differential gene are excluded to reduce noise. Clustering is based on euclidean distance in order to generate p-values via an approximately unbiased method with bootstrap probabilities at each node. Clusters are defined when the derived p-value is above the threshold of alpha greater than 0.95. After cluster assignment, the GDS in each sub-classification cluster is used for PubMed literature retrieval. To check for batch effects during clustering, the number of samples, date of submission, and platform type for each GDS is collected. Each cluster is compared to the overall distribution for a significant difference (ɑ < 0.01) using the Kolmogorov-Smirnov test and chi-squared test for numerical and categorical data, respectively. Batch statistics are output to text file.
A pairwise distance graph showing the GDS datasets based on absoulte difference in number of genes per GDS is also produced. Based on the gene sets defined above.
For more information on the function of PhenoX, please see our recent bioRxiv publication.
Lucy Lu Wang
Huaiying Lin
Agnes Bao
Subhajit Sengupta
Ben Busby
Robert R Butler III