GDDP - Computational genetic disease diagnosis based on phenotypes

Welcome to GDDP, computational Genetic Disease Diagnosis based on Phenotypes.

It's very easy to query for OMIM diseases based on phenotypes using this application:

Select "Query" from the top navigation bar
Paste or enter the phenotype text
Select the computational method (Method2 is selected by default)
Click Submit
Select "HPO terms" for recognized HPO terms and "OMIM diseases" for ranked OMIM disease list

Statistics of current disease knowledge base:

Text HPO

Select method

SS cutoff

Remove redundancy

Show plot parameters

Please enter phenotype

Download results

May 11, 2023 Web service API completed.

May 5, 2023 Annotation updated: Number of OMIM diseases: 8232. Number of HPO Phenotypic Abnormality terms: 10247.

February 1, 2021 Annotation updated: Number of OMIM diseases: 7871. Number of HPO Phenotypic Abnormality terms: 8038.

April 10, 2019 Annotation updated: Number of OMIM diseases: 7196. Number of HPO Phenotypic Abnormality terms: 7816.

August 2, 2018 Application migration and deployment to dedicated VM server.

June 12, 2018 Paper published online.

February 28, 2018 Web application v1.0 completed.

September 5, 2017 Web application v0.0 completed.

Janunary 23,2017 Annotation updated: Number of OMIM diseases: 7036. Number of HPO Phenotypic Abnormality terms: 7934.

Chen J, Xu H, Jegga A, Zhang K, White P, Zhang G. 2018. Novel phenotype-disease matching tool for rare genetic diseases. Genetics in Medicine. [PDF]

Method 1: integrated semantic similarity

Evaluating similarities for all pairs of phenotype terms between query and reference database.
calculating a similarity score to summarize the similarities between all the query terms (Q) and the HPO terms annotated to a target disease D_k.

Method 2: weighted overlapping

In this method, the phenotypes of a patient (query terms, Q) and HPO terms annotated to diseases (D_k) are first ‘up-induced’ based on HPO tree structure so that if a HPO term is annotated to a patient/disease, all of its ancestors are also annotated to the patient/disease. In order to compare the query terms (Q) with the terms annotated to a disease (D_k), we can construct a weighted 2x2 contingency table that contains the weighted counts of HPO terms shared or not shared between the query terms and the terms annotated to a disease. A Fisher’s exact test is then applied to this 2x2 contingency table and the p-value from the test can be used to rank the concordance/discordance between the query terms and the phenotypes of a disease.

Base URL

Endpoints

text2hpo: Map phenotype description in free text (input: text [string]) to HPO terms.
hpo2omim: Rank OMIM diseases based on a list of HPO IDs (input: hpo [string]) using GDDP Method 2.
hpo2gene: Rank OMIM genes based on a list of HPO IDs (input: hpo [string]) using GDDP Method 2.