eccCL

eccCL implements GPU-accelerated Ensemble Classifier Chains (ECC) to perform high-throughput multi-label classification for bioinformatics applications such as protein function prediction and drug resistance testing in HIV.

Key Features:

• Parallelized GPU implementation: Leverages GPUs to parallelize Classifier Chains (CC) and Ensemble Classifier Chains (ECC) computations for accelerated multi-label classification.
• High-throughput performance: Demonstrates processing speeds exceeding 25,000 instances per second on GPU hardware.
• OpenCL and R-package support: Provides an OpenCL implementation for cross-platform GPU execution and an R-package for integration into R-based workflows.
• Optimization for large datasets: Optimizes algorithms specifically for GPU execution to handle large volumes of data typical of next-generation sequencing experiments.
• Prediction accuracy: Adapting ECC for GPU execution preserves or can enhance prediction accuracy compared with CPU implementations.

Scientific Applications:

• Biomedical research: Enhances protein function prediction and supports drug resistance testing in HIV through multi-label classification.
• Genomics and bioinformatics: Facilitates analysis of extensive genomic datasets generated by next-generation sequencing technologies.

Methodology:

Adapts the Ensemble Classifier Chains framework for GPU execution via OpenCL, parallelizing computations across GPU hardware.