DrugMatrix ApicalXOmics

DrugMatrix Database

The DrugMatrix database is an integrated toxicogenomic that contains the short-term rat toxicity study results from approximately 600 pharmaceuticals and environmental chemicals performed. The complete DrugMatrix database (plain text link: https://cebs.niehs.nih.gov/cebs/paper/15670) is available on NIEHS CEBS site .

Briefly:

~ 700 Short-term toxicity studies (0.25 to 5 days) in male SD rats
~ 637 compounds studied at multiple doses, time points and tissues
~ 4,000 dose-time-tissue combinations (biological triplicates)
~ 13,000 Codelink RU1 Microarrays
~ 5,000 Affymetrix RG230-2 Arrays
~ 127,000 histopathology measurements
~ 100,000 hematology and chemistry measurements
~ 130 different in vitro assays
~ 900 chemicals with detailed literature curation

Search Strategy

A combination of endpoints were accessed in each study including target organ histopathology, clinical chemistry and target organ toxicogenomics. This design allows for derivation of relationships between the different endpoint (e.g. identification of transcriptional biomarkers of pathology).

To this end we have created a Shiny web application on top of the DrugMatrix database that allows users to

query a gene and identify its relationship to all diagnosed pathologies, clinical pathologies, and experimental animal organ weight changes due to the chemial compound treatment
query a specific pathology, clinical pathology, organ weight change to identify the most strongly associated genes.

Users can refine the search by selecting the criteria of duration of exposure, organ/tissue source of gene expression, gene probe, histopathology, etc., on a microarray platform (Codelink and/or Affymetrix) and then click the SUBMIT button. Results can be visualized graphically and are downloadable in multiple formats.

Quick Guide to Perform a Search

In addition to the current Project Description page, we have six other tabs:

Genes to Pathology

If you are interested in knowing which histopathology is most significantly affected in a specific tissue organ, you choose a tissue and an array chip (Coldlink RU1 or Affymetrix RG230) with an exposure time (5 days).
Within about a minute or so, you will be able to see the data retrieved from the DrugMatrix database. The significance is measured by looking at the 3 columns in the Summary table displayed on the main panel - gene expression level average Log10 Ratio DIFF, T-value, and P-value (see explanation below).
Once you see the top row displayed in the Summary table, you may click on the pathology name, e.g. TUBULE, REGENERATION. You will see all the experiments associated with the choice in detail shown in the table under the Summary table. In the meantime, a box plot with overlaying scatter plot shows at the bottom of the main panel.
To generate a report, you may choose PDF, csv, or Excel file format as shown in the upper corner of the table of your interest.

Pathology to Genes

If you are interested in knowing which gene is most significantly involved or associated in a specific tissue organ, you choose a tissue and an array chip with an exposure time (5 days). Note: Since there are ~ 34K gene probes needed to scan through, the query takes much longer up to ~4 minutes.
Similarly, the significance is measured by looking at the 3 columns in the Summary table displayed on the main panel, i.e. Log10 Ratio DIFF, T-value, and P-value.
Once you see the top row displayed in the Summary table, you may click on the gene probe name, e.g. Havcr1. You will see all the experiments associated with the choice in detail shown in the table under the Summary table. In the meantime, a list of associated pathological images, if there are any, with various severities (normal, minimal, mild, moderate, and marked) show on top of the main panel. A box plot with overlaying scatter plot shows on the bottom of the page. You may download the tabular data report the same way as described above.

Genes to Clinical Pathology

You select a gene of interest to see which clinical assay is strongly expressed in an organ/tissue selected. Assay effect was determined by comparing the averaged assay value to the range of normal. Namely, if the average value falls in between the range, it is considered to be normal. Hence, higher than the upper bound of the range, it is increase. Lower than the lower bound, it is decrease.

Clinical Pathology to Genes

You select a clinical assay and an assay effect, i.e. (increase or decrease ) as described above to see which gene expression is more significantly associated.

Genes to Organ Weight Change

You select a gene of interest with the direction of organ/tissue weight change to see which expression source, e.g. LIVER, is highly associated with organ weight change.

Organ Weight Change to Genes

You choose a direction of organ/tissue weight change along with the gene expression tissue source, to find out which gene is mosted involved in the association with the weight change. It is noted that the cut-off criteria about organ/tissue weight change is decided by 10% organ weight increase or decrease relative to the mean of whole body weight.

Analysis applied with gene association:

DIFF - the Log10 ratio difference between severity score at zeros (control group) and the greater than zeros. Click the column header to sort.
T-value - value generated from paired observations between treatment group of average Log10 Ratio and non-treatment group of average Log10 Ratio.
P-value - calculated with Mann-Whitney method for the treatment group severity scores greater than zeros, range of normal, and level of gene expression. Values ≥ 0.05 are highlighted in pink.

In the interface, boxplots, together with dotted plots are used to visualize the distribution quickly to examine for outliers. You can compare the multiple groups on the same variable or multiple variables on the same scale. To illustrate this, here is an example. If you look at the histopathology TUBULE REGENERATION in Genes to Pathology tab, the boxplots are created to compare the chemical compound treatments with zero severity scores versus the the treatments with severity score ≥ 0s. Each boxplot represents the expression levels (i.e. log10 ratio) across the severities. A dotted plot overlapped on the boxplot shows that each dot represents an individual experimental treatment at a concentration within a specific time exposure (e.g. 5 days).

Exemplary Search Output

Take an example output from Pathology to Genes tab. It shows the gene experssion levels denoted by average log10 ratios versus various severity scores when choosing Abcc3 gene probe in Liver. Above the Summary table, you see a list of pathological images assocaited with the liver.

Click on the histopathology name on the row with severity, you get the pathological image shown as above.

You may download a tabular output data in the format of some choices such as PDF, csv, or excel, etc.