Journal: bioRxiv

Article Title: VaLPAS: Leveraging variation in experimental multi-omics data to elucidate protein function

doi: 10.64898/2026.03.26.712966

Figure Lengend Snippet: A) The mean confidence (positive predictive value) for the top 1000 predicted associations calibrated by KEGG module membership is shown for each omics type (rows) and each association type (columns). These results show that the novel AI method provides the best performance for annotating proteins. B) The overlap of associations from autoencoder application to each omics type is shown on left. On right is the overlap of the corresponding proteins/genes involved. This shows that each omics type contributes annotations to different subsets of molecules.

Article Snippet: The potential of this approach has long been recognized in the analysis of high-throughput omics data ( ; ), but does not have a standard approach or framework to explore and characterize these associations.

Techniques:

Journal: Scientific Reports

Article Title: Characterizing the omics landscape based on 10,000+ datasets

doi: 10.1038/s41598-025-87256-5

Figure Lengend Snippet: Data characteristics of datasets investigated in this study. The data characteristics were calculated on log2-transformed data. ‘mtx’ denotes the entire quantitative matrix of the dataset, where i refers to analytes (rows) and and j to samples (columns). NIPALS = nonlinear iterative partial least squares, PCA = principal component analysis.

Article Snippet: The general concept that there is no one-fits-all data processing strategy is also applicable for biological high-throughput (omics) data.

Techniques: Standard Deviation