Artificial intelligence driven multi-omics framework identifies COL6A3 as a diagnostic biomarker and a putative gene target modulated by Embelin in Colorectal cancer

TL;DR

This study uses AI and multi-omics to identify COL6A3 as a potential biomarker and target for Embelin in colorectal cancer treatment.

Contribution

The novel integration of AI and multi-omics identifies COL6A3 as a biomarker and a target modulated by Embelin in colorectal cancer.

Findings

COL6A3 was identified as a hub gene by three ML algorithms with high accuracy (AUC > ~0.90).

Embelin treatment modulates apoptosis-related genes, upregulating BAX and CASPASE3 while downregulating BCL2 and PI3KCA.

Abstract

The third leading cause of death worldwide is colorectal cancer due to a lack of early detection biomarkers and therapeutic small molecules. Advances in systems biology offer a combination of multi-omics and Artificial intelligence to discover the potential biomarkers and targets.

We used a combination of in silico and in vitro methodologies to identify potential biomarkers and a putative mediator of Embelin in colon cancer treatment. The human colorectal cancer (gene expression profiling by array) datasets were analyzed by using Weighted Gene Co-expression Analysis (WGCNA), and predictive AI models were trained by three algorithms (LASSO, SVM-RFE, RF). All three algorithms predicted COL6A3 as a common hub gene. qRT-PCR was used to analyze the expression level of COL6A3 along with apoptosis markers in HCT116 cell lines (human colorectal cancer) by treating Embelin in a dose-dependent manner.

Trained model predicted COL6A3 as a prominent hub gene across all three ML algorithms with high cross validation accuracy (AUC values: > ~0.90), showing the accuracy of predictions and feature selections of the trained model. Embelin treatment results in the upregulation of pro-apoptotic markers (BAX, CASPASE3) and the downregulation of anti-apoptotic genes (BCL2, PI3KCA). These findings suggest that COL6A3 is a candidate biomarker and a potential mediator of embelin activity.

This study underscores the integration of AI, multi-omics, and in vitro studies for the discovery of candidate biomarkers and mechanistic insights into pathway modulation by Embelin in colorectal cancer. The research successfully identified and validated the role of COL6A3 as a potential biomarker and putative target modulated by Embelin in colon cancer.

A flowchart illustrating a research process with three main steps: 1. Co-expression network construction features a laptop showing a network diagram. 2. Machine-learning models involve a database icon and training models with graphs. 3. In vitro validations depict a petri dish labeled HCT116, a chemical structure for Embelin treatment, and an image of COL6A3 interacting with ECM. Targets are predicted, and gene expression analysis is conducted using a machine with a display showing percentage data.

Figures