Skip to content

S-LDSC Analysis of RVEF

I used Stratified Linkage Disequilibrium Score Regression (S-LDSC) to analyze summary statistics from a GWAS of MRI-determined right ventricular ejection fraction (RVEF) by Pirruccello et al.1

I used the standard reference data sources to associate chromatin regions with cell or tissue types. These reference data sources were:

Results

GTEx and Franke lab tissue expression data

The plot below illustrates the pattern of S-LDSC p values across cell types. The table shows the cell types with the lowest p values.

gene-expression-sldsc-rvef

Name Coefficient Coefficient_P_value Reject Null
Heart_Left_Ventricle 2.59059e-08 4.37654e-05 True
Muscle_Skeletal 2.28636e-08 0.000112161 False
A07.541.560.Heart.Ventricles 2.25644e-08 0.000183698 False
A07.541.Heart 2.16562e-08 0.000203794 False
Heart_Atrial_Appendage 1.91676e-08 0.000272105 False
A07.541.358.Heart.Atria 1.68989e-08 0.00406107 False
A02.633.567.850.Quadriceps.Muscle 1.82694e-08 0.00554677 False
A07.541.358.100.Atrial.Appendage 1.5286e-08 0.00909596 False

Given that our phenotype is related to the contractive power of a ventricle, this makes sense. I assume that the match with left ventricular tissue type is caused by transcriptomic overlap between the left and right ventricles

It is also interesting to observe that all the top cell types are musculoskeletal. This indicates that the key expression patterns that LDSC is picking up are likely characteristic of muscle.

Roadmap Chromatin data

I next applied S-LDSC to the RVEF GWAS using reference data generated by Finucane et al.2 from the Roadmap Epigenetics Project

The results are shown in the graph and table below.

lvef_ldsc_chromatin_plot

Name Coefficient Coefficient_P_value Reject Null
Right_Ventricle__H3K27ac 3.94707e-07 1.33824e-06 True
Left_Ventricle__H3K27ac 2.80259e-07 1.34429e-06 True
Fetal_Heart__H3K9ac 3.62558e-07 2.31896e-06 True
Stomach_Smooth_Muscle__H3K4me3 6.11652e-07 2.7594e-06 True
Fetal_Heart__DNase 3.67744e-07 5.20833e-06 True
Right_Atrium__H3K27ac 3.42679e-07 1.7163e-05 True
Colon_Smooth_Muscle__H3K4me3 6.20742e-07 7.11116e-05 True
Skeletal_Muscle_Male__H3K4me3 5.99152e-07 0.000103465 True
Ovary__H3K4me1 2.59535e-07 0.000340226 False
Fetal_Heart__H3K4me1 1.3078e-07 0.000356224 False
Ovary__H3K27ac 3.01364e-07 0.000510812 False
Colon_Smooth_Muscle__H3K27ac 1.74983e-07 0.000570302 False
Esoph-GJ_ENTEX__H3K4me3 5.49898e-07 0.00060134 False
Heart-Atrial_ENTEX__H3K27ac 1.24243e-07 0.000604863 False
Colon_Smooth_Muscle__H3K4me1 1.71125e-07 0.000623398 False
Fetal_Muscle_Leg__H3K4me3 9.45169e-07 0.00067477 False
Skeletal_Muscle_Female__H3K4me3 4.72043e-07 0.00123306 False
NHDF-Ad_Adult_Dermal_Fibroblast_Primary_Cells__H3K36me3 4.26679e-07 0.00133146 False

Impressively, the most significant tissue here is the right ventricle, consistent with the phenotype. Moreover, all other significant tissues are types of muscle, reinforcing the impression that LDSC has picked up on muscle-specific gene expression patterns.

The presence of "Ovary" at a low but insignificant p value may be noise, or may point to a sex-hormone effect on heart function.

ImmGen Data

No ImmGen cell types were significant, consistent with RVEF being a mainly non-immune phenotype.

Here are the lowest p values:

Name Coefficient Coefficient_P_value Reject Null
Tgd.vg4+24ahi.e17.Th 2.82104e-08 0.00539746 False
Tgd.vg2+24ahi.e17.Th.v2 2.13139e-08 0.00648692 False
CD4.24h.LN 2.13795e-08 0.010458 False
B.Fo.LN 1.39012e-08 0.0130979 False
B1a.PC 1.20947e-08 0.0233794 False
CD8.24h.LN 1.64771e-08 0.0254857 False
DC.pDC.8+.Sp 1.2325e-08 0.0278243 False
LEC.SLN.CFA.d6.v2 1.25117e-08 0.0288681 False
preB.FrD.BM 1.15162e-08 0.0325545 False
B.Fo.MLN 1.17467e-08 0.0338449 False
preB.FrD.FL 1.72978e-08 0.035241 False
Tgd.vg2+24ahi.Th 1.4413e-08 0.038868 False

Corces et al. ATAC-seq data

Similar to the above, no cell types were significant when we used the Corces ATAC-seq dataset. The lowest p values are below

Name Coefficient Coefficient_P_value Reject Null
Erythro 2.35709e-07 0.119 False
Bcell 4.14389e-08 0.337901 False
MEP -3.19943e-08 0.632292 False
MPP -4.04893e-08 0.694066 False
HSC -4.9625e-08 0.733438 False
CMP -6.57733e-08 0.768703 False
CLP -1.05289e-07 0.789269 False
GMP -1.24871e-07 0.931607 False

Cahoy and GTEx-Brain data

No cell types were significant in the Cahoy or GTEx brain datasets, indicating that the GWAS did not detect significant associations between neurological cell types and right ventricular function.

Name Coefficient Coefficient_P_value Reject Null
Neuron 8.00207e-09 0.124024 False
Oligodendrocyte -6.04124e-09 0.823426 False
Astrocyte -5.98411e-09 0.867185 False
Name Coefficient Coefficient_P_value Reject Null
Brain_Cerebellar_Hemisphere 1.10584e-08 0.10896 False
Brain_Cerebellum 7.64012e-09 0.196354 False
Brain_Frontal_Cortex_(BA9) 8.0182e-10 0.434653 False
Brain_Cortex 2.58682e-10 0.48244 False
Brain_Substantia_nigra 3.19984e-11 0.497729 False
Brain_Hippocampus -3.92956e-10 0.520903 False
Brain_Anterior_cingulate_cortex_(BA24) -4.1335e-10 0.534662 False
Brain_Hypothalamus -1.52283e-09 0.613854 False
Brain_Amygdala -1.59616e-09 0.620004 False
Brain_Spinal_cord_(cervical_c-1) -4.26574e-09 0.754433 False
Brain_Caudate_(basal_ganglia) -6.2817e-09 0.92293 False
Brain_Putamen_(basal_ganglia) -1.20152e-08 0.986481 False
Brain_Nucleus_accumbens_(basal_ganglia) -1.4452e-08 0.992423 False

Reproducing

To reproduce this analysis, use the right heart analysis script.

  1. James P Pirruccello, Paolo Di Achille, Victor Nauffal, Mahan Nekoui, Samuel F Friedman, Marcus DR Klarqvist, Mark D Chaffin, Lu-Chen Weng, Jonathan W Cunningham, Shaan Khurshid, and others. Genetic analysis of right heart structure and function in 40,000 people. Nature Genetics, 54(6):792–803, 2022. URL: https://pmc.ncbi.nlm.nih.gov/articles/PMC10313645/. ↩

  2. Hilary K Finucane, Yakir A Reshef, Verneri Anttila, Kamil Slowikowski, Alexander Gusev, Andrea Byrnes, Steven Gazal, Po-Ru Loh, Caleb Lareau, Noam Shoresh, and others. Heritability enrichment of specifically expressed genes identifies disease-relevant tissues and cell types. Nature Genetics, 50(4):621–629, 2018. URL: https://pmc.ncbi.nlm.nih.gov/articles/PMC5896795/. ↩