Heterogeneous Cloud Framework for Big Data Genome Sequencing.

Area of Science:

• Bioinformatics
• Computational Biology
• Big Data Analytics

Background:

• Next-generation genome sequencing generates massive datasets, challenging current computational infrastructures.
• Existing acceleration methods often rely on single approaches, proving insufficient for explosive data scales and complexities.
• Efficient processing of short and long reads is crucial for advancing genomic research.

Purpose of the Study:

• To propose and evaluate a novel FPGA-based acceleration solution integrated with the MapReduce framework for next-generation genome sequencing.
• To address the limitations of current methodologies in handling large-scale genomic data.
• To enhance the speed and efficiency of aligning short reads to reference genomes.

Main Methods:

• Development of a novel FPGA-based acceleration architecture utilizing the MapReduce framework on multiple hardware accelerators.
• Theoretical speedup analysis conducted on a MapReduce programming platform.
• Implementation and evaluation of a hardware prototype on a Xilinx FPGA chip.

Main Results:

• Theoretical analysis demonstrated significant potential for speedup in large-scale genome sequencing applications.
• Experimental evaluation on an FPGA prototype showed efficient acceleration of the next-generation sequencing problem.
• The proposed platform achieved satisfactory accuracy, mapping quality, and acceptable hardware cost with a low error rate.

Conclusions:

• The proposed FPGA-based MapReduce solution effectively accelerates next-generation genome sequencing tasks.
• This approach offers a viable and efficient method for handling the computational demands of modern genomics.
• The system provides a balance of performance, accuracy, and cost-effectiveness for genomic data analysis.