DNA Sequence Classification

Classifying coding and non-coding regions using CNN

Project Overview

The DNA Sequence Classification project focuses on training a Convolutional Neural Network (CNN) to distinguish between coding and non-coding regions in DNA sequences. Using datasets from Ensembl and NCBI, the project leverages PyTorch, BioPython, and various Python libraries for data processing, model training, and visualization.

Objectives

  • Classify DNA sequences into coding and non-coding regions.
  • Build and train a CNN using PyTorch.
  • Visualize data distributions and model performance.

Tools & Libraries

  • PyTorch – Deep learning framework
  • BioPython – DNA sequence processing
  • Pandas & NumPy – Data manipulation
  • Matplotlib – Data visualization

Data Sources

Detailed Process

  1. Data Preprocessing
    • Extracted DNA sequences from FASTA and GenBank files.
    • Applied one-hot encoding and K-mer frequency encoding.
  2. Model Training
    • Designed a 1D CNN for sequence classification.
    • Used cross-entropy loss and optimized using Adam.
  3. Evaluation & Results
    • Measured accuracy, precision, recall, and F1-score.
    • Visualized confusion matrix and ROC curve.

Key Insights

  • The CNN achieved 92% accuracy in distinguishing coding regions.
  • Data augmentation techniques improved generalization.

For the full code and documentation, visit the GitHub Repository.

This enhanced project page includes a detailed overview, key insights, and direct links, presenting your work effectively.

References