Hyojung Paik

Relating Chemical Structure to Cellular Response: An Integrative Analysis of Gene Expression, Bioactivity, and Structural Data Across 11,000 Compounds

A central premise in systems pharmacology is that structurally similar compounds have similar cellular responses; however, this principle often does not hold. One of the most widely used measures of cellular response is gene expression. By integrating gene expression data from Library of Integrated Network‐based Cellular Signatures (LINCS) with chemical structure and bioactivity data from PubChem, we performed a large‐scale correlation analysis of chemical structures and gene expression profiles of over 11,000 compounds taking into account confounding factors such as biological conditions (e.g., cell line, dose) and bioactivities. We found that structurally similar compounds do indeed yield similar gene expression profiles. There is an ∼20% chance that two structurally similar compounds (Tanimoto Coefficient ≥ 0.85) share significantly similar gene expression profiles. Regardless of structural similarity, two compounds tend to share similar gene expression profiles in a cell line when they are administrated at a higher dose or when the cell line is sensitive to both compounds.

Repurpose terbutaline sulfate for amyotrophic lateral sclerosis using electronic medical records.

Prediction of new disease indications for approved drugs by computational methods has been based largely on the genomics signatures of drugs and diseases. We propose a method for drug repositioning that uses the clinical signatures extracted from over 13 years of electronic medical records from a tertiary hospital, including >9.4 M laboratory tests from >530,000 patients, in addition to diverse genomics signatures. Cross-validation using over 17,000 known drug–disease associations shows this approach outperforms various predictive models based on genomics signatures and a well-known “guilt-by-association” method. Interestingly, the prediction suggests that terbutaline sulfate, which is widely used for asthma, is a promising candidate for amyotrophic lateral sclerosis for which there are few therapeutic options. In vivo tests using zebrafish models found that terbutaline sulfate prevents defects in axons and neuromuscular junction degeneration in a dose-dependent manner. A therapeutic potential of terbutaline sulfate was also observed when axonal and neuromuscular junction degeneration have already occurred in zebrafish model. Cotreatment with a β2-adrenergic receptor antagonist, butoxamine, suggests that the effect of terbutaline is mediated by activation of β2-adrenergic receptors.

Comprehensive analysis of normal adjacent to tumor transcriptomes

Histologically normal tissue adjacent to the tumor (NAT) is commonly used as a control in cancer studies. However, little is known about the transcriptomic profile of NAT, how it is influenced by the tumor, and how the profile compares with non-tumor-bearing tissues. Here, we integrate data from the Genotype-Tissue Expression project and The Cancer Genome Atlas to comprehensively analyze the transcriptomes of healthy, NAT, and tumor tissues in 6506 samples across eight tissues and corresponding tumor types. Our analysis shows that NAT presents a unique intermediate state between healthy and tumor. Differential gene expression and protein–protein interaction analyses reveal altered pathways shared among NATs across tissue types. We characterize a set of 18 genes that are specifically activated in NATs. By applying pathway and tissue composition analyses, we suggest a pan-cancer mechanism of pro-inflammatory signals from the tumor stimulates an inflammatory response in the adjacent endothelium.Normal tissue adjacent to the tumour (NAT) is often used as a control in cancer studies. Here, the authors analyse across cancer types the transcriptomes of healthy, NAT, and tumour tissues, and find that NAT presents a unique state, potentially due to inflammatory response of the NAT to the tumour tissue.

Unraveling human protein interaction networks underlying co-occurrences of diseases and pathological conditions.

BackgroundHuman diseases frequently cause complications such as obesity-induced diabetes and share numbers of pathological conditions, such as inflammation, by dysfunctions of common functional modules, such as protein–protein interactions (PPIs).MethodsOur developed pipeline, ICod (Interaction analysis for disease Comorbidity), grades similarities between pairs of disease-related PPIs including comorbid diseases and pathological conditions. ICod displayed a disease similarity network consisting of nodes of disease PPIs and edges of similarity value. As a proof of concept, eight complex diseases and pathological conditions, such as type 2 diabetes, obesity, inflammation, and cancers, were examined to discover whether PPIs shared between diseases were associated with comorbidities.ResultsBy comparing Medicare reports of disease co-occurrences from 31 million patients, the disease similarity network shows that PPIs of pathological conditions, including insulin resistance, and inflammation, overlap significantly with PPIs of various comorbid diseases, including diabetes, obesity, and cancers (p < 0.05). Interestingly, maintaining connectivity between essential genes was more drastically perturbed by removing a node of a disease-related gene rather than a pathological condition-related gene, such as one related to inflammations.ConclusionThus, PPIs of pathological symptoms are underlying functional modules across diseases accompanying comorbidity phenomena, whereas they contribute only marginally to maintaining interactions between essential genes.

Reversal of cancer gene expression correlates with drug efficacy and reveals therapeutic targets

The decreasing cost of genomic technologies has enabled the molecular characterization of large-scale clinical disease samples and of molecular changes upon drug treatment in various disease models. Exploring methods to relate diseases to potentially efficacious drugs through various molecular features is critically important in the discovery of new therapeutics. Here we show that the potency of a drug to reverse cancer-associated gene expression changes positively correlates with that drug’s efficacy in preclinical models of breast, liver and colon cancers. Using a systems-based approach, we predict four compounds showing high potency to reverse gene expression in liver cancer and validate that all four compounds are effective in five liver cancer cell lines. The in vivo efficacy of pyrvinium pamoate is further confirmed in a subcutaneous xenograft model. In conclusion, this systems-based approach may be complementary to the traditional target-based approach in connecting diseases to potentially efficacious drugs.

Creating ethnicity-specific reference intervals for lab tests from EHR data

The results of clinical lab tests are an essential component of medical decision-making. To guide interpretation, test results are returned with reference intervals defined by the range in which 95% of values occur in healthy individuals. Clinical laboratories often set their own reference intervals to accommodate local population and instruments variations. This approach is costly and can be biased. We describe a novel data-driven method for using electronic health record data to extract healthy patients’ information to define reference intervals. We found that the distributions of many clinical lab tests differ among self-identified racial and ethnic groups (SIREs) in healthy patients. Finally, we derived SIRE-specific reference intervals and provide evidence that these intervals have clinical prognostic value. Specifically, we show that for two lab tests, serum creatinine level and hemoglobin A1C, SIRE-specific reference intervals are more predictive for need for dialysis and development type 2 diabetes than existing reference intervals. One Sentence Summary A novel method for defining population-specific reference intervals of common clinical laboratory tests from electronical health records has better prognostic value than existing reference intervals.

Precision annotation of digital samples in NCBI’s gene expression omnibus

The Gene Expression Omnibus (GEO) contains more than two million digital samples from functional genomics experiments amassed over almost two decades. However, individual sample meta-data remains poorly described by unstructured free text attributes preventing its largescale reanalysis. We introduce the Search Tag Analyze Resource for GEO as a web application (http://STARGEO.org) to curate better annotations of sample phenotypes uniformly across different studies, and to use these sample annotations to define robust genomic signatures of disease pathology by meta-analysis. In this paper, we target a small group of biomedical graduate students to show rapid crowd-curation of precise sample annotations across all phenotypes, and we demonstrate the biological validity of these crowd-curated annotations for breast cancer. STARGEO.org makes GEO data findable, accessible, interoperable and reusable (i.e., FAIR) to ultimately facilitate knowledge discovery. Our work demonstrates the utility of crowd-curation and interpretation of open ‘big data’ under FAIR principles as a first step towards realizing an ideal paradigm of precision medicine.

Integrating Clinical Phenotype and Gene Expression Data to Prioritize Novel Drug Uses

Drug repositioning has been based largely on genomic signatures of drugs and diseases. One challenge in these efforts lies in connecting the molecular signatures of drugs into clinical responses, including therapeutic and side effects, to the repurpose of drugs. We addressed this challenge by evaluating drug‐drug relationships using a phenotypic and molecular‐based approach that integrates therapeutic indications, side effects, and gene expression profiles induced by each drug. Using cosine similarity, relationships between 445 drugs were evaluated based on high‐dimensional spaces consisting of phenotypic terms of drugs and genomic signatures, respectively. One hundred fifty‐one of 445 drugs comprising 450 drug pairs displayed significant similarities in both phenotypic and genomic signatures (P value < 0.05). We also found that similar gene expressions of drugs do indeed yield similar clinical phenotypes. We generated similarity matrixes of drugs using the expression profiles they induce in a cell line and phenotypic effects.