CCIDB: a manually curated cell-cell interaction database with cell context information

Abstract

Cell-cell interaction (CCI) is a crucial event in the development and function of multicellular organisms. The development of CCI databases is beneficial for researchers who want to analyze single-cell sequencing data or study CCI through molecular experiments. CCIs are known to act differently according to cellular and biological contexts such as cell types, gene mutations or disease status; however, previous CCI databases do not completely provide this contextual information pertaining to CCIs. We constructed a cell-cell interaction database (CCIDB) containing the biological and clinical contexts involved in each interaction. To build a database of cellular and tissue contexts, we collected 38 types of context features, which were categorized into seven categories, including 'interaction', 'cell type', 'cofactor', 'effector', 'phenotype', 'pathology' and 'reference'. CCIs were manually retrieved from 272 studies published recently (less than 6 years ago). In the current version of CCIDB, 520 CCIs and their 38 context features have been manually collected and curated by biodata engineers. We suggest that CCIDB is a manually curated CCI resource that is highly useful, especially for analyzing context-dependent alterations in CCIs. Database URL https://ccidb.sysmed.kr/.

Figure 1.

Distribution of CCIDB. (a–d) Pie plots showing the distribution of the features in CCIDB, including species (a), disease type (b), cancer type (c), interaction type (d, left) and signaling type of the LR interactions (d, right). (e–f) Bar plots showing the top-ranked 10 frequent source cell types (e, left), target cell types (e, right), source genes (f, left) and target genes (f, right).

Figure 2.

Intracellular communication network analysis using CCIDB in breast cancer. (a) Cell types of the 100,064 cells of breast cancer patients are indicated in a UMAP plot. (b) The inferred intracellular communication network across cell types. Circle size is proportional to the number of cells in each cell group and edge width represents the communication probability. (c) A Venn diagram shows the distribution of interaction pairs identified from CCIDB and CellChatDB. (d) A circle plot shows the interactions of top ranked source-target gene pairs across cell groups. The circle color and size represent the calculated communication probability and P-values, respectively. The red asterisks indicate the pairs found in CCIDB.

Figure 3.

Construction of CCI network. (a) The liver cancer network (LCN) was constructed with six context features, namely ‘source tissue’, ‘source cell type’, ‘source gene’, ‘target gene’, ‘target cell type’ and ‘target tissue’ using Cytoscape. (b) The LCN (left) and BCN (right) constructed by integrating CCIDB and previous four CCI databases are shown. The source databases of CCIDB (blue) and others (cyan) are indicated in different colors, and hub pair genes are indicated (red).