[데이터과학연구센터 세미나] 곽일엽 교수님 (중앙대학교 응용통계학과)
■ 세미나 안내
일시: 2025년 11월 25일 (화) 오후4시
장소: 경영대학1호관 308호 세미나실
강사: 곽일엽 교수님 (중앙대학교 응용통계학과)
강연제목: From Competition to Contribution: Cell Lineage Reconstruction and Gene Expression Prediction through DREAM Challenges
초록:
In this talk, I will share our experience participating in two DREAM challenges, which evolved into interesting research contributions. First, in the "Allen Institute Cell Lineage Reconstruction" challenge, we tackled the complex task of reconstructing single-cell lineage trees from CRISPR recorders. By introducing novel methods like Weighted Hamming Distance (WHD) and K-mer Replacement Distance (KRD), we achieved top performance in Subchallenges 2 and 3, providing robust lineage predictions validated across various scenarios. These methods demonstrated robust performance across various conditions and are now available as the DCLEAR R package for public use. Second, in the "Predicting Gene Expression using Random Promoter Sequences" challenge, we developed Proformer, a hybrid macaron transformer model that predicts gene expression values from promoter sequences. I will explain the step-by-step process our team followed to design and refine this model. Our model secured 3rd place in the competition, and I will also introduce how we collaborated with the 1st and 2nd place teams, as well as the organizers, to derive the optimal solution through joint research. Through these experiences, we demonstrate how participation in scientific challenges bridges competitive innovation with collaborative progress, paving the way for advances in gene regulation modeling and single-cell lineage research.

다음글