umbilical vein endothelial cell line  (ATCC)

Journal: medRxiv

Article Title: Machine Learning and Micro Capture-C resolve GWAS associations revealing endothelial stress pathways in Coronary Artery Disease

doi: 10.64898/2025.12.18.25342557

Figure Lengend Snippet: Machine Learning predicts the effect of genetic variants in coronary artery cell types. A. REnformer was trained on publicly available scATAC data from primary coronary artery tissue samples. REnformer was then used to predict the genome accessibility for 9 coronary artery cell types. An example locus around the MMP2 gene is shown. The tracks show the ground truth scATAC data (labelled “ATAC” followed by the cell type), and directly below the REnformer prediction for that cell type. B. The effect on genome accessibility of 21,348 CAD associated lead SNPs and high linkage disequilibrium candidate causal SNPs is predicted by REnformer across 9 coronary artery cell types as well as HUVECs. Predicted coverage (Y-axis) shows the predicted chromatin accessibility at a given SNP position. Alt-Ref prediction (X-axis) shows the predicted effect size of the candidate CAD SNP. Empirical significance cut-offs of 0.01 were calculated for each cell type; variants below the cut-off (loss of function) are coloured blue, variants above the cut-off (gain of function) are coloured red. C. Example REnformer predictions at the chr1:q25.1 locus for SNP rs1879454 (reference allele = C, alternate allele =A). The position of rs1879454 is shown by a blue triangle. D. The prediction difference for rs1879454 where the reference allele has been subtracted from the alternate allele. E. Feature attribution for the REnformer predictions in panel C. The relative contribution of each base pair in the reference and alternate genome to the prediction is shown by height (positive or negative) of the relevant letter. The variant position is shown by a blue triangle.

Article Snippet: Single donor human umbilical vein endothelial cell (HUVEC) lines were purchased from Promocell (C-12200) or were isolated from umbilical cord tissue samples provided by the Anthony Nolan Trust, who received ethical approval from the East Midlands-Derby, UK Research Ethics Committee and written, informed consent from the donor’s parents.

Techniques: Variant Assay

Journal: medRxiv

Article Title: Machine Learning and Micro Capture-C resolve GWAS associations revealing endothelial stress pathways in Coronary Artery Disease

doi: 10.64898/2025.12.18.25342557

Figure Lengend Snippet: A. Genome regulatory elements including enhancers (Enh), CTCF sites, and promoters (Pr) were determined for HUVECs using REgulamentary. B. Chart showing the number of intersections between REgulamentary defined regulatory elements and candidate causal variants.

Article Snippet: Single donor human umbilical vein endothelial cell (HUVEC) lines were purchased from Promocell (C-12200) or were isolated from umbilical cord tissue samples provided by the Anthony Nolan Trust, who received ethical approval from the East Midlands-Derby, UK Research Ethics Committee and written, informed consent from the donor’s parents.

Techniques:

Journal: medRxiv

Article Title: Machine Learning and Micro Capture-C resolve GWAS associations revealing endothelial stress pathways in Coronary Artery Disease

doi: 10.64898/2025.12.18.25342557

Figure Lengend Snippet: REnformer prioritises candidate causal variants in HUVECs. REnformer predictions are shown for 4 example candidate causal variants: rs604723 (A), rs2508619 (B), rs582384 (C), rs17293632 (D). In each panel, the prioritised SNP is labelled above the gene annotation, and other non-prioritised SNPs are shown as vertical black lines. The REnformer prediction (alt-ref) is shown below the gene annotation and directly above ATACseq data for HUVECs. Publicly available ChIPseq tracks (H3K4me3, H3K4me1, H3K27ac, CTCF) are also shown to aid identification of the genome regulatory elements. REnformer feature attribution for the 40bp around the prioritised SNP (yellow highlight) is shown below the genome tracks in each panel; the reference allele feature attribution is shown above the alternate allele. Statistically significant transcription factor binding motifs are shown below the feature attribution in B, C, and D. The identity of the transcription factor binding the affected motif in A is unknown.

Article Snippet: Single donor human umbilical vein endothelial cell (HUVEC) lines were purchased from Promocell (C-12200) or were isolated from umbilical cord tissue samples provided by the Anthony Nolan Trust, who received ethical approval from the East Midlands-Derby, UK Research Ethics Committee and written, informed consent from the donor’s parents.

Techniques: Binding Assay

Journal: medRxiv

Article Title: Machine Learning and Micro Capture-C resolve GWAS associations revealing endothelial stress pathways in Coronary Artery Disease

doi: 10.64898/2025.12.18.25342557

Figure Lengend Snippet: Analysis of the endothelial enhancer-promoter map for CAD genetics. A. String diagram showing genes linked by MCC to a HUVEC enhancer containing a candidate causal CAD SNP. Interactions between a captured MCC enhancer and gene promoters were called using the bespoke MCC peakcaller MCC_LoTron. Gene promoters were defined using REgulamentary. Nodes which are enriched for specified GO terms are highlighted by colour. The thickness of the connecting line corresponds to the strength of the interaction evidence between two nodes. B. Gene Ontology analysis (biological processes enrichment) of the MCC derived gene network.

Article Snippet: Single donor human umbilical vein endothelial cell (HUVEC) lines were purchased from Promocell (C-12200) or were isolated from umbilical cord tissue samples provided by the Anthony Nolan Trust, who received ethical approval from the East Midlands-Derby, UK Research Ethics Committee and written, informed consent from the donor’s parents.

Techniques: Derivative Assay

Journal: medRxiv

Article Title: Machine Learning and Micro Capture-C resolve GWAS associations revealing endothelial stress pathways in Coronary Artery Disease

doi: 10.64898/2025.12.18.25342557

Figure Lengend Snippet: POINTseq detects the changes in MCC linked output genes. A. Schematic overview of experimental approach. MCC links candidate causal enhancer SNPs to the promoters of output genes. Long read sequencing is used to phase SNPs and reconstruct the alleles from the donor cells. ATACseq and POINTseq data is re-mapped to the reconstructed alleles and reads are assigned to the allele they are derived from based on corresponding SNPs. The amount of signal coming from each allele can then be measured and compared, demonstrating the effect of a casual SNP in an enhancer on gene transcriptional output. B. Example POINTseq data tracks compared to traditional RNAseq. POINTseq, polyA-, and polyA+ RNAseq were performed on 3 HUVEC donors. Reads are split between forward and reverse DNA strand, mapped to the hg38 reference genome, and displayed as bigwig files. POINTseq offers far greater transcriptional gene coverage as introns are retained. D. Plot of POINTseq measured allelic skew in genes linked by MCC to candidate causal enhancer SNPs. Direction of change is displayed with respect to the reference allele versus the alternate allele (x-axis). Significance (log10 p-value) is displayed on the y-axis. Detected significant genes are labelled.

Article Snippet: Single donor human umbilical vein endothelial cell (HUVEC) lines were purchased from Promocell (C-12200) or were isolated from umbilical cord tissue samples provided by the Anthony Nolan Trust, who received ethical approval from the East Midlands-Derby, UK Research Ethics Committee and written, informed consent from the donor’s parents.

Techniques: Sequencing, Derivative Assay