Biomarker search method, biomarker search device, and program

A device for searching a biomarker to be used for determining whether or not a first treatment has an effect on a disease, the biomarker search device comprising:

• a first storage storing clinical information of a plurality of patients before treatment;
• a second storage storing omics information of a plurality of patients before treatment;
• a third storage storing omics information of control;
• a fourth storage storing information indicating the effect of treatment in a plurality of patients;
• a signature molecular information detector that detects signature molecular information peculiar to disease from the omics information comprising a plurality of molecular information items, based on the omics information of patients and the omics information of control;
• a regression analyzer that carries out a regression analysis using the effect of treatment as an objective variable, and using the clinical information before treatment and variable data of at least one molecular information item selected as molecular information to be processed from the signature molecular information as explanatory variables;
• a network generator that generates a network structure based on partial correlation between a first node, a second node and a third node, the first node being the effect of treatment, the second node being the clinical information before treatment, and the third node being the selected molecular information to be processed as the nodes, when the determination coefficient of the regression analysis by the regression analyzer is equal to or more than a first threshold value;
• and
• a biomarker determinator that determines whether or not there is a causal relationship equal to or more than a second threshold value between the molecular information to be processed and the effect of treatment in the network structure, and determining the molecular information to be processed as a biomarker, when there is a causal relationship equal to or more than the second threshold value.

A non-transitory computer-readable medium storing a program for searching a biomarker to be used for determining whether or not a first treatment has an effect on a disease, the program making a computer execute:

• reading, from a storage, omics information of a plurality of patients before treatment and omics information of control;
• detecting signature molecular information peculiar to disease from the omics information comprising a plurality of molecular information items, based on the read omics information of patients and the read omics information of control;
• and
• obtaining a biomarker from the omics information of the signature molecular information, wherein
• upon obtaining the biomarker,
• selecting at least one molecular information item to be processed from the signature molecular information,
• reading the clinical information of a plurality of patients before treatment and the information of the effect of treatment in a plurality of patients from a storage,
• carrying out a regression analysis using the effect of treatment as an objective variable, and using the clinical information before treatment and variable data of the molecular information to be processed as explanatory variables,
• generating a network structure based on partial correlation between a first node, a second node and a third node, the first node being the effect of treatment, the second node being the clinical information before treatment, and the third node being the molecular information to be processed, when the determination coefficient of the regression analysis is equal to or more than a first threshold value, and
• a process of determining whether or not there is a causal relationship equal to or more than a second threshold value between the molecular information to be processed and the effect of treatment in the network structure, and
• determining the molecular information to be processed as a biomarker, when there is a causal relationship equal to or more than the second threshold value
• are repeated.