Conceptual Approach to Clustering in the Study of Gene Expression

Abstract

The work contains an overview of clustering methods used in the study of gene expression and new results of application of Formal Concepts Analysis to clustering data extracted from the public functional genomics data repository GEO. Methods of Formal Concept Analysis allow clustering of multidimensional data under the single condition of partial ordering of such data sets. As a result, clusters are separate sublattices in the concept lattice, where each sublattice contains hierarchically related formal concepts. This solution of the clustering problem allows deep investigating both mutual influences of genes and their influence on other data obtained in experiments on gene expression. The paper describes a new information technology developed for the implementation of the proposed approach. The technology uses modern solutions in the field of Big Data processing, it has functions for communication with external data sources and other information systems. Preliminary results of the application of this technology to three-dimensional gene expression data obtained from the GEO system are presented.

Characterization of Traffic Accidents Based on Long-Horizon Aggregated and Disaggregated Data

“…Clustering analysis is a mathematical statistical method that can be applied to large datasets, where the raw data are sorted and grouped into clusters [30][31][32][33][34][35][36][37][38]. These clusters are internally homogeneous, although they are different from each other.…”

“…The difference between the two methodologies is that multidimensional scaling identifies the key dimensions, while clustering analysis identifies the groups. Clustering analysis is considered to be the opposite of factor analysis [33]. Factor analysis reduces the number of variables by grouping them into smaller groups of factors, but clustering analysis reduces the number of observations or cases by grouping them into smaller groups [37].…”

“…However, refs. [32,33] seriously criticized Smeed's model, as data from only one year were utilized in the model's development. They also pointed out that Smeed's model could not be used for all countries because each country has distinct traffic, economic, and social parameters, and that the model's coefficient and exponent should, therefore, be country dependent.…”

Clustering Methods in Gene Expression Research

“…Анализ формальных понятий (АФП) [ Любые данные экспрессии генов могут быть представлены в виде формальных контекстов. На формальных контекстах строятся решётки понятий, представляющие собой решения задач би- [22] и трикластеризации [23], а также мультимодальной кластеризации любой размерности. При этом формально не используется оптимизация и кластеризацию можно считать точной.…”