Skip to content

S-LDSC RBCV Analysis

I applied Stratified Linkage Disequilibrium Score Regression1 (S-LDSC) to summary statistics from a GWAS of Red Blood Cell Volume from the Million Veterans Program.

I used standard S-LDSC reference datasets. These reference datasets were originally derived from:

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.

s-ldsc_rcvol_expression

Name Coefficient Coefficient_P_value Reject Null
A15.145.300.Fetal.Blood 8.02657e-08 6.59384e-09 True
A15.145.Blood 5.94179e-08 7.3992e-06 True
A11.443.Erythroid.Cells 4.41781e-08 1.19728e-05 True
A11.627.635.Myeloid.Progenitor.Cells 4.62524e-08 0.000107559 True
A11.872.378.Hematopoietic.Stem.Cells 4.41011e-08 0.000172671 True
A15.382.Immune.System 3.99963e-08 0.000342322 False
Whole_Blood 3.64886e-08 0.000399347 False
A15.145.229.188.Blood.Platelets 3.14462e-08 0.000464739 False
A11.872.378.590.817.Megakaryocyte.Erythroid.Progenitor.Cells 3.67321e-08 0.000603583 False
A15.145.229.Blood.Cells 4.13485e-08 0.000654442 False
A11.118.637.555.567.562.B.Lymphocytes 3.57229e-08 0.000800263 False
Spleen 3.89864e-08 0.000854701 False
A11.118.637.555.567.569.T.Lymphocytes 2.90771e-08 0.00104616 False
Colon_Transverse 2.6992e-08 0.00126636 False
A15.145.229.637.555.567.569.200.CD4.Positive.T.Lymphocytes 2.53835e-08 0.00230767 False
A15.382.490.555.567.537.Killer.Cells..Natural 3.36356e-08 0.00263635 False
A11.627.340.360.Granulocyte.Precursor.Cells 2.73892e-08 0.00301885 False
A15.145.229.637.555.567.562.725.Plasma.Cells 3.18685e-08 0.0035067 False
A11.118.637.Leukocytes 3.64806e-08 0.00398084 False
A15.382.520.604.800.Palatine.Tonsil 2.93966e-08 0.00419334 False
A11.118.637.555.567.569.200.700.T.Lymphocytes..Regulatory 2.38386e-08 0.00485846 False
A11.872.Stem.Cells 2.63215e-08 0.00487292 False
A11.118.637.555.567.562.440.Precursor.Cells..B.Lymphoid 2.5128e-08 0.00587332 False
A11.872.378.590.635.Granulocyte.Macrophage.Progenitor.Cells 2.81929e-08 0.00629756 False
A03.556.124.684.Intestine..Small 2.71434e-08 0.0065321 False
A15.382.490.555.567.Lymphocytes 3.1885e-08 0.00745577 False
A03.556.249.124.Ileum 2.81618e-08 0.00763379 False
Cells_EBV-transformed_lymphocytes 2.55655e-08 0.00789262 False

The 5 significant cell types are: "A15.145.300.Fetal.Blood", "A15.145.Blood", "A11.443.Erythroid.Cells", "A11.627.635.Myeloid.Progenitor.Cells".

These findings makes sense, and are consistent with known biology: "Erethryoid cells" are a class that includes standard red blood cells (erythrocytes), while myeloid progenitor cells are ancestors to red blood cells (see the figure below).

haemopoetic-cells

Figure: By A. Rad and M. Häggström. CC-BY-SA 3.0 license

Roadmap Chromatin data

I next applied S-LDSC to the red blood cell volume GWAS using reference data generated by Finucane et al.1 from the Roadmap Epigenetics Project.

The following graph and table show the results:

s-lsdc-rbcvol-chromatin

Name Coefficient Coefficient_P_value Reject Null
Primary_hematopoietic_stem_cells_G-CSF-mobilized_Female__H3K27ac 9.7986e-07 4.30062e-07 True
Primary_hematopoietic_stem_cells_G-CSF-mobilized_Female__H3K4me1 3.89211e-07 7.36713e-07 True
Primary_hematopoietic_stem_cells_G-CSF-mobilized_Female__DNase 9.67532e-07 1.22959e-06 True
Primary_hematopoietic_stem_cells__H3K4me1 7.08867e-07 3.25765e-06 True
Primary_hematopoietic_stem_cells_G-CSF-mobilized_Male__H3K4me1 4.06036e-07 3.37892e-06 True
Primary_T_killer_memory_cells_from_peripheral_blood__H3K27ac 8.61697e-07 3.55327e-06 True
Primary_hematopoietic_stem_cells_short_term_culture__H3K4me3 1.833e-06 5.35886e-06 True
Primary_hematopoietic_stem_cells_G-CSF-mobilized_Male__DNase 8.90935e-07 7.23909e-06 True
Primary_T_helper_memory_cells_from_peripheral_blood_2__H3K27ac 6.6463e-07 2.52533e-05 True
Primary_hematopoietic_stem_cells_G-CSF-mobilized_Male__H3K4me3 1.0486e-06 2.63718e-05 True
Spleen_ENTEX__H3K27ac 2.05768e-07 4.87077e-05 True
Fetal_Thymus__H3K36me3 2.79975e-07 5.17104e-05 True
Primary_hematopoietic_stem_cells_short_term_culture__H3K4me1 3.5475e-07 8.47972e-05 True
Primary_hematopoietic_stem_cells_G-CSF-mobilized_Female__H3K4me3 1.75499e-06 0.000127767 True
Primary_hematopoietic_stem_cells_short_term_culture__H3K36me3 3.44274e-07 0.000129613 True
Spleen_ENTEX__H3K4me3 3.49278e-07 0.000195307 True
Primary_neutrophils_from_peripheral_blood__H3K4me3 6.33005e-07 0.000318123 True
Primary_B_cells_from_peripheral_blood__DNase 4.68288e-07 0.000512658 False
Primary_neutrophils_from_peripheral_blood__H3K4me1 1.87677e-07 0.000658363 False
Spleen_ENTEX__H3K4me1 9.59291e-07 0.00107238 False
Rectal_Mucosa_Donor_31__H3K9ac 5.6792e-07 0.00116076 False
Stomach_ENTEX__H3K4me3 7.71536e-07 0.00122671 False
SI-Term-Ileum_ENTEX__H3K4me3 6.86199e-07 0.00140399 False
Primary_hematopoietic_stem_cells__H3K4me3 1.4126e-06 0.00167475 False
Primary_T_helper_cells_from_peripheral_blood__H3K4me3 7.52953e-07 0.00177749 False
Fetal_Thymus__DNase 4.10651e-07 0.00233247 False
SI-Term-Ileum_ENTEX__H3K27ac 2.38673e-07 0.0023326 False
Primary_Natural_Killer_cells_from_peripheral_blood__H3K4me1 2.12516e-07 0.00384186 False

Again, consistent with known biology, the significant cell types are all in the blood/immune category and related to or descended from haematopoietic stem cells.

The presence of lymphocytes like T-cells on the list of significant cell types is interesting. My guess is that this may reflect epigenetic features shared by all cells in haematopoietic stem cell lineage, rather than true etiological role for lymphocytes in red blood cell volume.

ImmGen Data

The next step is to use the S-LDSC reference data derived from the ImmGen project.

immgen-rbc-plot

Name Coefficient Coefficient_P_value Reject Null
preT.DN2.Th 7.01741e-08 7.55644e-08 True
T.8SP69+.Th 6.10773e-08 4.08943e-07 True
preT.ETP-2A.Th 6.13855e-08 4.64764e-07 True
preT.DN2-3.Th 5.88185e-08 5.87392e-07 True
T.4int8+.Th 5.54558e-08 6.74419e-07 True
preT.DN2A.Th 5.66926e-08 1.21757e-06 True
Tgd.vg2+24ahi.Th 5.37253e-08 2.51879e-06 True
CD4.96h.LN 5.33706e-08 4.68263e-06 True
T.8SP24int.Th 4.98821e-08 4.85374e-06 True
Tgd.Th 5.3534e-08 6.5081e-06 True
preT.DN3-4.Th 5.45315e-08 7.09887e-06 True
preT.DN3B.Th 5.52977e-08 1.07887e-05 True
SC.LT34F.BM 4.19728e-08 1.54079e-05 True
Tgd.vg2-.Sp 4.27974e-08 1.98276e-05 True
preT.ETP.Th 5.19913e-08 2.82895e-05 True
T.8SP24-.Th 4.95349e-08 3.34413e-05 True
T.DP.69-.e17.Th.v2 4.53702e-08 3.40748e-05 True
proB.FrBC.FL 4.98927e-08 3.65938e-05 True
MF.BM 6.94982e-08 3.83549e-05 True
SC.MDP.BM 4.9775e-08 3.98549e-05 True
proB.CLP.FL 4.93602e-08 4.49107e-05 True
T.4Nve.Sp 4.0749e-08 4.70371e-05 True
CD4.48h.LN 5.13317e-08 4.90148e-05 True
Tgd.vg2+24ahi.e17.Th 3.98931e-08 5.05495e-05 True
T.4+8int.Th 3.81905e-08 5.49732e-05 True
T.DP69+.Th.v2 5.14471e-08 5.7412e-05 True
NKT.44-NK1.1-.Th 3.6643e-08 6.27774e-05 True
T.4SP69+.Th 4.45522e-08 6.57873e-05 True

These results are somewhat surprising: all categories of immune-cell are tagged as significant. Moreover, T-helper cells are highly significant. My guess is that this again reflects shared features of cells descended from haematopoietic stem cells.

Corces et al. ATAC-seq data

Looking at the Corces et al. ATAC-seq dataset, we again see broad cross cell-type significance.

Name Coefficient Coefficient_P_value Reject Null
Erythro 2.20621e-06 5.0373e-11 True
MEP 5.39479e-07 4.68085e-08 True
CMP 3.95866e-07 7.26445e-05 True
HSC 3.3574e-07 0.000577361 True
MPP 2.90813e-07 0.00059752 True
Bcell 3.08234e-07 0.00107073 True
NK 3.10655e-07 0.00186229 True
GMP 2.39349e-07 0.00305697 True

However, here we also note that erythrocyte cells are by far the most significant, followed by their direct ancestors the Megakaryocytic-Erythroid Progenitors (MEP). This is consistent with the idea that erythrocytes are the cell type truly driving the phenotype, and the other cell types are significant to the extent that they share regulatory pathways with erythrocytes.

Cahoy and GTEx-Brain data

Name Coefficient Coefficient_P_value Reject Null
Astrocyte 8.37248e-09 0.137796 False
Oligodendrocyte 9.32556e-10 0.452566 False
Neuron -4.38565e-10 0.526099 False
Name Coefficient Coefficient_P_value Reject Null
Brain_Anterior_cingulate_cortex_(BA24) 1.51946e-08 0.00662582 False
Brain_Cerebellar_Hemisphere 1.18609e-08 0.0269974 False
Brain_Putamen_(basal_ganglia) 1.1835e-08 0.0664447 False
Brain_Cerebellum 7.09048e-09 0.164615 False
Brain_Nucleus_accumbens_(basal_ganglia) 5.70585e-09 0.170502 False
Brain_Cortex 6.52218e-09 0.180797 False
Brain_Caudate_(basal_ganglia) 4.90205e-09 0.237016 False
Brain_Amygdala 3.25693e-09 0.272557 False

None of the neurological cell types in these datasets are significant. This is consistent with RBC volume being a non-neurological phenotype.

Reproducing

To reproduce these results, run the analysis script here.

  1. 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/. ↩↩