서울대학교 생물정보연구소와 생물정보학 협동과정 공동 주최로 세미나를 아래와 같이 열고자 하오니, 많은 참여 바랍니다.

일시: 2018년 5월 17일 목요일 오후 5시

연사: 안태진(한동대학교)

장소: 서울대학교 25동 411호

TITLE: Deep learning based identification of cancer using gene expression data

ABSTRACT: Deep learning has proven outstanding performance in recognition and classification problems. Cancer has different gene expression profile when it is compared to normal. As many publically available cancer and normal gene expression data is available for now, it is plausible to address how far machine can recognize the cancer. We integrated gene expression data from Gene Expression Omnibus (GEO), The Cancer Gene Atlas (TCGA), accumulated 11,838 (Cancer: 6,446, Normal: 5,392) and 11,574 gene expression data from 26 primary sites, respectively.

Our fundamental challenge is how let the machine discriminate cancer and normal. Further challenge is how let the human interpret the rationale what machine learned for prediction. We suggest visual representation of gene sets which can be used not only in training the machine, but also in human interpretation of machine learning result.

Deep Neural Network (DNN) was able to identify cancer by the averaged 5-fold accuracy of 0.981 for the training set, 0.962 for the test set. Ensemble approach combining trained model from different gene set yield 0.995. We also analyzed contribution of individual genes to a patient’s probability of being classified as cancer. Most contributing genes are functionally annotated as ‘cell to cell communication’ and enriched at the ‘extra cellular space’. This finding is not much different to common sense of biology. Thus, we want to suggest DNN not only as an effective identification tool for cancer or normal, but also it might be an important tool to find genes which can characterize cancer.

EDUCATION AND TRAINING

B.S. Department of Bioscience and Food engineering, Handong University, South Korea (1995-1999)

M.S. Division of Molecular and Life Sciences, POSTECH(Pohang University of Science and Technology), South Korea (1999-2001)

Ph.D. Bioinformatics program, Seoul National University, South Korean (2008-2014)

1999.3 ~ 2001.2 : BRIC, 포항공과대학교, 포항, 대한민국

2001.3 ~ 2001.7 : 3rd Millenium Inc., 보스턴, 미합중국

2001.8 ~ 2017.8 : 삼성전자, 수원, 대한민국

2017.9 ~ 현재 : 한동대학교, 포항, 대한민국

SELECTED PUBLICATION

Kim SY, Ahn T, Bang H, et al. Acquired resistance to LY2874455 in FGFR2-amplified gastric cancer through an emergence of novel FGFR2-ACSL5 fusion. Oncotarget. 2017;8(9):15014-15022.

Lee JY, Park K, Lee E, et al. Gene Expression Profiling of Breast Cancer Brain Metastasis. Sci Rep. 2016;6:28623.

Kim J-Y, Ahn T, Im Y-H, et al. Prognostication of HER family gene expression collaborate with ESR1 expression in patients with triple negative breast cancer. Cancer Res 2016;76(4 Suppl):Abstract nr P2-08-19

Srikanth MR, Ahn T; Park T et al. Estimating functional coupling between cancer gene sub-networks using novel interaction measures. International Journal of Data Mining and Bioinformatics (IJDMB), 2016:5

Park K, Choi MK, Jung HH, et al. Molecular characterization of patients with pathologic complete response or early failure after neoadjuvant chemotherapy for locally advanced breast cancer using next generation sequencing and nCounter assay. Oncotarget. 2015;6(27):24499-24510.

Kim ST, Ahn T, Lee E, et al. Exploratory biomarker analysis for treatment response in KRAS wild type metastatic colorectal cancer patients who received cetuximab plus irinotecan. BMC Cancer. 2015;15:747.

Ahn T, Park T. Pathway-driven discovery of rare mutational impact on cancer. Biomed Res Int. 2014;2014:171892.

Ahn T, Lee E, Huh N, Park T. Personalized identification of altered pathways in cancer using accumulated normal tissue data. Bioinformatics. 2014;30(17):i422-429.

서울대학교 생물정보연구소

생물정보학 협동과정 공동주최