Identification of Disease Critical Genes Using Collective Meta-heuristic Approaches: An Application to Preeclampsia

Area of Science:

• Computational life sciences
• Bioinformatics
• Genomics

Background:

• Identifying disease-critical genes from large gene expression datasets is a significant computational challenge.
• Preeclampsia is a serious pregnancy complication affecting maternal and fetal health.

Purpose of the Study:

• To apply and compare meta-heuristic algorithms for identifying disease-critical genes associated with preeclampsia.
• To introduce and evaluate novel hybrid algorithms, HBMO-kNN and HS-kNN, for improved gene identification.

Main Methods:

• Four meta-heuristic algorithms were employed: Honey Bee Mating Optimization (HBMO), Harmony Search (HS), Differential Evolution (DE), and Genetic Algorithm (GA).
• Two novel hybrid algorithms, HBMO-kNN and HS-kNN, were developed using the k-nearest neighbor (kNN) classifier.
• Performance was evaluated by comparing these with DE-kNN and SGA-kNN on three different datasets.

Main Results:

• Classification accuracy for the meta-heuristic algorithms ranged from 92.46% to 100% across datasets.
• The proposed HBMO-kNN algorithm demonstrated the best performance with 99.64-100% accuracy on most datasets.
• DE-kNN achieved the second-best performance, with accuracies between 99.42% and 100%.

Conclusions:

• The identified disease-critical genes largely align with known preeclampsia-related genes.
• The HBMO-kNN and DE-kNN hybrid approaches are effective for identifying critical genes in complex diseases like preeclampsia.
• This study advances computational methods for gene expression data analysis in disease research.