Related Experiment Video
Updated: Jun 11, 2025
11:02
Detecting Somatic Genetic Alterations in Tumor Specimens by Exon Capture and Massively Parallel Sequencing
Published on: October 18, 2013
19.4K
Cancer detection with various classification models: A comprehensive feature analysis using HMM to extract a
Vijay Kalal1, Brajesh Kumar Jha1
1Department of Mathematics, School of Technology, Pandit Deendayal Energy University, Raysan, Gandhinagar, Gujarat 382007, India.
Computational Biology and Chemistry
|October 8, 2024
Summary
This study uses Hidden Markov Models (HMM) to identify unique DNA patterns in cancer cells, improving early detection and diagnosis. Machine learning models effectively distinguish malignant from non-malignant sequences, aiding cancer research.
Area of Science:
- Genomics
- Bioinformatics
- Computational Biology
Background:
- Genomic sequences contain vital information for cellular processes in both healthy and malignant cells.
- Identifying distinct nucleotide patterns is crucial for understanding cancer's molecular basis and enabling early diagnosis.
- Early cancer detection significantly impacts treatment efficacy and patient outcomes.
Purpose of the Study:
- To develop a novel feature extraction method using Hidden Markov Models (HMM) for identifying complex genomic patterns.
- To differentiate nucleotide patterns specific to malignant and non-malignant cells for improved cancer detection.
- To reduce computational costs in nucleotide prediction through efficient feature extraction and selection.
Main Methods:
- Utilized Hidden Markov Models (HMM) for feature extraction and selection from genomic sequences.
- Analyzed nucleotide patterns in both coding (CDS) and non-coding (NCDS) regions of DNA.
- Implemented and evaluated machine learning classifiers: Gradient-Boosted Decision Trees (GBDT), Random Forests (RF), Decision Trees (DT), and Support Vector Machines (SVM).
Main Results:
- Decision Trees (DT) and ensemble methods (RF, GBDT) demonstrated significant ability to differentiate malignant from non-malignant DNA sequences.
- Support Vector Machines (SVM) with appropriate kernels substantially enhanced cancer detection accuracy.
- The combined approach of HMM-based feature reduction and nucleotide pattern classification yielded improved performance and reliable cancer detection.
Conclusions:
- Hidden Markov Models provide an effective strategy for feature extraction in genomic sequence analysis.
- Machine learning classifiers, particularly SVM and ensemble methods, are powerful tools for accurate cancer detection based on DNA patterns.
- Integrating feature reduction with HMM-based classification offers a robust framework for reliable cancer identification and diagnosis.
