Speaker:
Lifeng Yan (HIMIS)
Time:
October 29 (Thu), 2026, 2:30-3:30 pm
Venue:
HIMIS 7A1
Title:
From String Matching to Heterogeneous Computing: Accelerating Sequence Alignment at Scale
Abstract：
Sequence alignment can be viewed as string matching with substitutions, insertions, and deletions, yet the scale of modern sequencing data makes it computationally demanding. Starting with dynamic programming and the Smith–Waterman algorithm, this talk introduces the seed–chain–extend approach to narrowing the search space. Drawing on my research on CPU–GPU heterogeneous platforms, I will discuss how algorithm restructuring, data organization, and coordinated scheduling accelerate sequence alignment. I will conclude with broader lessons for scientific computing: how algorithms can adapt to modern hardware, and how to balance computational work, data movement, and parallelism. No background in bioinformatics or GPU programming is assumed.