Related Experiment Video
Updated: Aug 15, 2026
10:40
Comprehensive Workflow for the Genome-wide Identification and Expression Meta-analysis of the ATL E3 Ubiquitin Ligase Gene Family in Grapevine
Published on: December 22, 2017
MLHeatmap: an interactive application for transcriptomic marker-panel discovery
Eun Young Lee1,2, Jihye Park1, Seoyeon Youn1
1Department of Biomedical Sciences, College of Bio-Convergence, Dankook University, Cheonan 31116, Republic of Korea.
Bioinformatics Advances
|August 14, 2026
Summary
MLHeatmap integrates transcriptomic data analysis for biomarker discovery, enabling compact gene panel identification. This tool streamlines normalization, classification, and visualization for efficient cancer subtype classification.
Area of Science:
- Bioinformatics
- Computational Biology
- Genomics
Background:
- Transcriptomic biomarker studies often require multiple, separate tools for distinct analytical steps.
- This fragmentation complicates workflows and hinders efficient marker-panel discovery.
Purpose of the Study:
- To develop MLHeatmap, a unified, cross-platform, browser-based workflow.
- To integrate normalization, classification, feature ranking, and visualization for transcriptomic data.
- To enable compact marker-panel discovery directly from count matrices.
Main Methods:
- MLHeatmap accepts count matrices and sample labels for comprehensive analysis.
- The workflow includes gene mapping, normalization, multiclass classification with nested cross-validation, and feature attribution.
- Supports multiple classifiers and panel-selection methods, including interactive heatmap visualization.
Main Results:
- Applied to colorectal cancer consensus molecular subtype (CMS) classification using TCGA data.
- Achieved 89.8% out-of-fold accuracy and 0.973 macro AUC with a 13-gene panel.
- External validation demonstrated robust performance in independent cohorts, with AUCs ranging from 0.895 to 0.970.
Conclusions:
- MLHeatmap provides an integrated solution for transcriptomic biomarker studies.
- Facilitates efficient discovery of compact, clinically relevant gene panels.
- Demonstrates high accuracy and generalizability in cancer subtype classification.
