gCUP

gCUP predicts HIV-1 coreceptor usage from next-generation sequencing (NGS) data using GPU-accelerated genotypic prediction models to enable high-throughput tropism inference.

Key Features:

• High Computational Efficiency: The model is optimized for GPU execution (demonstrated on NVIDIA GeForce GTX 460) and achieves classification speeds of over 175,000 sequences per second.
• Accuracy and Reliability: gCUP is built upon established genotypic prediction models and maintains the accuracy required for clinical diagnostics, having been rigorously tested and validated.
• Versatility and Adaptability: The underlying methodology can be adapted to other bioinformatics applications such as drug resistance prediction.

Scientific Applications:

• HIV Diagnostics: Enables rapid and accurate prediction of HIV coreceptor usage (tropism) from NGS data to support genotypic diagnostics.
• Research and Development: Supports large-scale analyses of sequence data for studies of HIV evolution, drug resistance mechanisms, and personalized medicine approaches.

Methodology:

The computational prediction model is parallelized and optimized specifically for GPU execution, harnessing GPUs to reduce the time required to analyze complex genetic data compared with traditional CPU-based computations.