Clustering analysis method and system for multi-omics data

Abstract

The embodiment of the invention discloses a clustering analysis method and system for multi-omics data, and the method comprises the steps: segmenting MR image information through employing a neural network, and extracting a high-throughput image hyper-parameter according to the segmentation information of each part; processing the clinical data, the demographic data and the laboratory inspection data to generate vector representations of different dimensions; performing multi-source data fusion on the high-throughput image data and the vector representations of different dimensions to obtain fused multi-source heterogeneous data; constructing a multi-source heterogeneous data set, and training and testing a multi-source graph clustering model to obtain an optimal model; inputting MR image information into the optimal model, and analyzing differences of different categories and similarities of the same category. A graph structure mode is adopted, the correlation condition between data is visually expressed, different features are captured, the model is more robust, the efficient clustering algorithm based on the graph neural network model is achieved, and the practical value is very high.

Description

technical field [0001] The embodiments of the present application relate to the technical field of data analysis, and in particular to a method and system for cluster analysis of multi-omics data. Background technique [0002] In recent years, multi-omics analysis has been widely used in medical subtype analysis, and high-throughput data fusion to accurately define subtypes has become an important topic in precision medicine. At present, the integration of multi-omics is facing many problems, such as complex data, sparse data, and heterogeneous data. Data fusion methods can be divided into similarity methods, dimensionality reduction methods, and statistics-based methods. [0003] The similarity-based method uses the similarity between samples to cluster data. Since the number of samples is far smaller than the number of features in the current multi-omics data, the similarity-based method is very difficult when the sample size is insufficient. It is difficult to have a be...

AI Technical Summary

Problems solved by technology

[0003] The similarity-based method uses the similarity between samples to cluster data. Since the number of samples is far smaller than the number of features in the current multi-omics data, the similarity-based method is very difficult when the sample size is insufficient. Hard to have a better effect

Algorithms based on spectral clustering do not use node path information, which affects the accuracy of clustering

Images