Publications
Detailed Information
Optimization and Machine Learning Algorithms for Condition-Specific Biological Network Construction and Analyses : 최적화 및 기계학습 알고리즘을 통한 조건 특이적 생물 네트워크의 구성과 분석
Cited 0 time in
Web of Science
Cited 0 time in Scopus
- Authors
- Advisor
- 김선
- Major
- 공과대학 컴퓨터공학부
- Issue Date
- 2018-08
- Publisher
- 서울대학교 대학원
- Description
- 학위논문 (박사)-- 서울대학교 대학원 : 공과대학 컴퓨터공학부, 2018. 8. 김선.
- Abstract
- Network bioinformatics has been successfully used to help reveal the complex mechanism
of cells in research and industrial domains of diverse fields such as medical
science, the pharmaceutical industry, biological science, and agriculture. In network
bioinformatics, myriad of studies relied on the static association between interacting
nodes to construct the network or to discover novel knowledge from networks. However,
biological associations are condition-specific, which means that they can change
dynamically depending on the status of the cell. Thus, the needs for algorithms to
address these technical challenges have been increased. In this context, I developed
stochastic optimization (SO) and machine learning (ML) algorithms with sequencing
data that can capture status of cells. The main focus of my work is on ensemble SO
and ML approaches with network bioinformatics.
The first work is miRNA-mRNA regulation inference algorithm called PlantMirnaT
to construct condition-specific miRNA-mRNA bipartite network. This is a challenging
problem since it needs to consider expression data of two different types,
miRNA and mRNA, and target relationship between miRNA and mRNA is not clear,
especially when microarray data is used. Fortunately, due to the low sequencing cost,
small RNA and RNA sequencing are routinely processed and it may be able to infer
regulation relationships more accurately. To fully leverage the power of sequencing
data, I proposed a parameterized miRNA-mRNA expression model based on a novel
idea named split-ratio, and utilized genetic algorithm and the quasi-newton method
to determine optimal model parameters.
The second work is to discover functional subnetworks of miRNA-mRNA regulation
network that work in specific biological condition. These subnetworks are named
functional miRNA-mRNA regulatory module (MRM). Mining functional MRM has
exponential time complexity, thus heuristic algorithm is needed to discover optimal
MRM set. My algorithm operates in two steps: 1) grouping and ordering the miRNAs
and mRNAs to build per sample matrices representing miRNA-mRNA regulations,
and 2) determining maximum sized modules from structured miRNA-mRNA matrices.
The third work is to use deep learning method for network bioinformatics. Deep
learning has shown a great potential to address the various learning problems. However,
deep learning technologies conventionally use grid-like structured data, thus
application of deep learning technologies to the classification of human disease subtypes
is yet to be explored. Recently, graph based deep learning techniques have
emerged, which becomes an opportunity to leverage analyses in network biology. I
propose a hybrid model, which integrates two key components 1) graph convolution
neural network (graph CNN) and 2) relation network (RN). I utilize graph CNN as
a component to learn expression patterns of cooperative gene community, and RN as
a component to learn associations between learned patterns. The proposed model is
applied to the synthetic dataset and PAM50 breast cancer subtype classification task,
the standard breast cancer subtype classification of clinical utility. In experiments of
both subtype classification and patient survival analysis, my algorithm achieve the
better result in both quantitative and qualitative perspectives than previous methods.
- Language
- English
- Files in This Item:
Item View & Download Count
Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.