High-Order Correlation Integration for Single-Cell or Bulk RNA-seq Data Analysis.

ABSTRACT: Quantifying or labeling the sample type with high quality is a challenging task, which is a key step for understanding complex diseases. Reducing noise pollution to data and ensuring the extracted intrinsic patterns in concordance with the primary data structure are important in sample clustering and classification. Here we propose an effective data integration framework named as HCI (High-order Correlation Integration), which takes an advantage of high-order correlation matrix incorporated with pattern fusion analysis (PFA), to realize high-dimensional data feature extraction. On the one hand, the high-order Pearson's correlation coefficient can highlight the latent patterns underlying noisy input datasets and thus improve the accuracy and robustness of the algorithms currently available

SUBMITTER: Tang H 

PROVIDER: S-EPMC6497731 | biostudies-literature | 2019

REPOSITORIES: biostudies-literature