ChemPert: mapping between chemical perturbation and transcriptional response for non-cancer cells

Abstract

Prior knowledge of perturbation data can significantly assist in inferring the relationship between chemical perturbations and their specific transcriptional response. However, current databases mostly contain cancer cell lines, which are unsuitable for the aforementioned inference in non-cancer cells, such as cells related to non-cancer disease, immunology and aging. Here, we present ChemPert (https://chempert.uni.lu/), a database consisting of 82 270 transcriptional signatures in response to 2566 unique perturbagens (drugs, small molecules and protein ligands) across 167 non-cancer cell types, as well as the protein targets of 57 818 perturbagens. In addition, we develop a computational tool that leverages the non-cancer cell datasets, which enables more accurate predictions of perturbation responses and drugs in non-cancer cells compared to those based onto cancer databases. In particular, ChemPert correctly predicted drug effects for treating hepatitis and novel drugs for osteoarthritis. The ChemPert web interface is user-friendly and allows easy access of the entire datasets and the computational tool, providing valuable resources for both experimental researchers who wish to find datasets relevant to their research and computational researchers who need comprehensive non-cancer perturbation transcriptomics datasets for developing novel algorithms. Overall, ChemPert will facilitate future in silico compound screening for non-cancer cells.

Figure 1.

Schematic outline of ChemPert. (A) The sources and three main components of ChemPert database. (B) Illustration of built-in algorithms in ChemPert. One option for predicting the TF responses given the perturbagen and expression profile of initial cellular state (Option 1) and the other for predicting perturbagens that induce desired transcriptional response (Option 2).

Figure 2.

The compositions and evaluation of ChemPert database. (A) Distribution of datasets across different cell types/lines/tissues in the ChemPert database. Y-axis scale is log₂(number + 1) for each cell type/line/tissue. (B) Frequency of TFs in the ChemPert database, including inhibited and activated ones. X-axis represents the frequency of TFs and y-axis presents the number of TFs with corresponding frequency. (C) Distribution of perturbagen frequency in the ChemPert database. X-axis represents the frequency of perturbagen, and y-axis represents the number of perturbagen with corresponding frequency. (D) Distribution of datasets for different perturbation durations. (E) AURPC for response TF prediction given perturbagens. (F) AURPC for protein target prediction given response TFs. (G) Number of datasets with correct perturbagen prediction, data are mean ± MSE. E–G used the benchmarking datasets to compare the performance of ChemPert tool using the ChemPert database, cancer database or randomization. Significance was calculated by using one-sided Wilcoxon test. ***: P-value < 2.22e−16. (H) Fraction of perturbagens whose within-ness are significantly larger than between-ness.

Figure 3.

Illustration of ChemPert web interface. (A) The home page of web interface. ChemPert mainly consists of two sections: database and webtool. (B) The home page of the ChemPert database. The database is composed of three parts: targets of perturbagens, gene expression of initial cell types and transcriptional response. Clicking the button can switch to corresponding part. (C) The webtool page. Users can predict either the response TFs of given perturbagen or the perturbagens targeting desired query TFs. (D) The transcriptional response table listing the meta information of datasets. (E) Detailed transcriptional response for one dataset. (F) Information about targets of perturbagens.

Figure 4.

Application of ChemPert. (A–F) Venn diagrams showing overlaps of predicted TFs among different diets and disease states of NASH models. Up-regulated TFs (A) and down-regulated TFs (B) after OCA perturbation, up-regulated TFs (C) and down-regulated TFs (D) after pioglitazone perturbation, up-regulated TFs (E) and down-regulated TFs (F) after vitamin E perturbation. (G) The representative of predicted perturbagens with literatures support for the treatment of OA. Details are shown in Supplementary Table S4.