Journal: Nature Communications

Article Title: MultiGATE: integrative analysis and regulatory inference in spatial multi-omics data via graph representation learning

doi: 10.1038/s41467-025-63418-x

Figure Lengend Snippet: a Bright-field image and manually annotated segmentation of hippocampus layers and white matter (WM) in the human hippocampus. b Spatial clustering of hippocampal regions using MultiGATE, SpatialGlue, and Seurat WNN. Clustering performance is assessed using the Adjusted Rand Index (ARI), with higher values indicating greater clustering accuracy. c Box plots representing attention scores for peak–gene pairs across different genomic distances, grouped based on whether they are supported by expression quantitative trait loci (eQTL) evidence. The box plots indicate the medians (centerlines), means (triangles), first and third quartiles (bounds of boxes), and 1.5 × interquartile range (whiskers). Sample sizes per bin (False/True): 0–25 kb (621/222), 25–50 kb (479/88), 50–75 kb (461/78), 75–100 kb (469/44), 100–125 kb (446/30), 125–150 kb (405/29). d Receiver operating characteristic (ROC) curves comparing the performance of MultiGATE and other methods in predicting eQTL-associated regulatory interactions. e Visualization of MultiGATE-predicted cis-regulatory interactions for the target genes CA12 and PRKD3 along with eQTL evidence. Source data are provided as a Source Data file.

Article Snippet: For the adult human hippocampus dataset (spatial ATAC–RNA–seq), the mouse brain dataset (spatial transcriptomics + metabolomics), the breast cancer-patterned spatial ATAC + RNA dataset (Supplementary Fig. ), which have ground truth spatial domain annotations, we computed the following metrics to assess clustering agreement: Rand Index (RI), ARI, Adjusted Mutual Information, Normalized Mutual Information, Homogeneity, Completeness, V-measure, Fowlkes-Mallows Index.

Techniques: Expressing

Journal: Journal of Proteome Research

Article Title: Automated Mag-Net Enrichment Unlocks Deep and Cost-Effective LC–MS Plasma Proteomics

doi: 10.1021/acs.jproteome.5c00420

Figure Lengend Snippet: Enriched and neat plasma proteins cluster into four main groups based on protein intensities. Cluster 1 contains proteins found in neat plasma and shared across all methods, while Cluster 2 consists of proteins identified in all five enrichment methods. Cluster 3 includes proteins with medium intensity, mostly detected in all methods except ENRICHiST and neat. Cluster 4 consists primarily of low-intensity proteins. Each cluster is enriched with distinct Gene Ontology terms, cellular localizations, and tissue associations. (A) Hierarchical clustering of proteins detected in at least 10% of all samples, based on log 2 intensity. (B) Enrichment analysis of proteins in each cluster with the PANTHER GO-Slim databases , for cellular components and biological processes. Results were first filtered to retain terms with over 2-fold enrichment and a minimum of ten proteins per term. The top five terms with the lowest FDR values were selected for each cluster. (C) Subcellular location of proteins in each cluster based on The Human Protein Atlas (HPA) annotations. Only the top 20 locations are shown. The inset highlights the top five secretome annotations among all identified proteins. (D) Number of proteins annotated as tissue-enriched in the HPA transcriptomics data set. The 20 tissues with the highest total count of tissue-enriched proteins are shown.

Article Snippet: When comparing our protein identification results to the Human Protein Atlas (HPA) transcriptomics data set, we found that the most tissue-enriched proteins were predominantly expressed in plasma cells/bone marrow and hepatocytes/liver.

Techniques: Clinical Proteomics, Transcriptomics