Network-Constrained Eigen-Single-Cell Profile Estimation for Uncovering Crucial Immunogene Regulatory Systems in
Heewon Park1, Satoru Miyano2,3
1School of Mathematics, Statistics and Data Science, Sungshin Women's University, Seoul, Republic of Korea.
Summary
We developed a new computational method to identify key markers for acute myeloid leukemia (AML) progression by analyzing cell line gene expression. This approach highlights CD79A as a critical factor in AML, offering insights into disease mechanisms.
Area of Science:
- Computational Biology
- Genomics
- Cancer Research
Background:
- Acute myeloid leukemia (AML) progression is linked to age-related cellular changes.
- Existing eigenCell analysis methods struggle with noise and interpretability due to dense feature combinations.
Purpose of the Study:
- To develop a novel computational approach for characterizing age-related phenotypes in AML cell lines.
- To identify key molecular markers associated with AML progression using sparse learning and network biology.
Main Methods:
- Developed network-constrained eigenCells profile estimation using sparse learning (LASSO and network penalization).
- Incorporated network biology to select hub and regulator/target genes for sparse eigenCell estimation.
- Applied Monte Carlo simulations to validate the method's efficacy in reconstructing sparse structures.
Main Results:
- Successfully estimated sparse eigenCell profiles, highlighting critical markers.
- Identified key markers for age-related phenotypes in healthy and AML cell lines, supported by existing literature.
- Unveiled the regulatory system of immunogenes, pinpointing CD79A subnetwork activity as pivotal in AML progression.
Conclusions:
- The proposed network-constrained eigenCells method effectively identifies disease-related markers by incorporating network biology insights.
- Diminished activity in the CD79A subnetwork is implicated as a significant mechanism in AML progression.
- This computational tool can aid in characterizing disease-specific cell line subsets, phenotypes, and clones.
