a, Negative correlation between PLT and MPV in a UK sample. The gender-adjusted correlation coefficient r and trend line are shown. b, Protein–protein interaction network of platelet loci. For the nodes, genes are represented by round symbols, where node colour reflects gene transcript level in megakaryocytes on a continuous scale from low (dark green) to high (white). Grey-coloured round symbols identify first-order interactors identified in Reactome and IntAct. Core genes not connected to the main network are omitted. The 34 core genes are identified by a blue perimeter. Yellow perimeters identify five additional genes (VWF, PTPN11, PIK3CG, NFE2 and MYB) with known roles in haemostasis and megakaryopoiesis and mapping to within the association signals at distances greater than 10 kb from the sentinel SNPs. These genes, which do not conform to the rule for inclusion into the core gene list, are not considered in further analyses presented in Fig. 2c, d and Supplementary Fig. 5 and are shown here for illustration purposes only. Network edges were obtained from the Reactome (blue) and IntAct-like (red) databases and through manual literature curation (black). The network including the 34 core genes alone contains 633 nodes and 827 edges; after inclusion of the 5 additional genes, the network (shown here) includes 785 nodes and 1,085 edges. The full network, containing gene expression levels and other annotation features, is available in Cytoscape²⁵ format for download (Supplementary Data 1). c, d, Time course experiments of gene expression in megakaryocytes and erythroblasts. Expression of core genes in log₂ transformed signal intensities (log₂ SI) during differentiation of the haematopoietic stem cells into megakaryocytes (c) or erythroblasts (d), segregated by their trends of statistically significant increasing (red), decreasing (blue) or unchanged (grey) gene expression. The corresponding gene list for the three classes is given in Supplementary Data 1.

Gene-specific morpholinos were injected into wild-type and Tg(cd41:EGFP) embryos at the one cell stage (Supplementary Fig. 6) to assess alterations in erythropoiesis and thrombopoiesis. a, Control D. rerio embryo at 72 h post fertilization (h.p.f.); the boxed region corresponds to images in the middle panels of b–h. b–h, Left: o-dianisidine staining was used to assess the number of mature erythrocytes at 48 h.p.f.: ehd3 (h) morpholino-injected embryos showed normal haemoglobin staining, whereas embryos injected with ak3 (c), rnf145 (d), arhgef3 (e) or jmjd1c (g) morpholinos showed a decrease in the number of haemoglobin-positive cells compared to control embryos (b). Embryos injected with tpma morpholinos (f) showed normal numbers of erythrocytes but unusual accumulation dorsally in the blood vessels (compatible with cardiomyopathy). Middle: haematopoietic stem-cell and thrombocyte development was assessed using the transgenic Tg(cd41:EGFP) line at 72 h.p.f. Embryos injected with the ehd3 (h) morpholino had a normal number of GFP¹ cells in the caudal haematopoietic tissue and circulation, when compared to control embryos (b). However, GFP¹ cells were absent in ak3 (c), rnf145 (d), arhgef3 (e), tpma (f) and jmjd1c (g) morpholino-injected embryos. Right: One-cell-stage embryos were injected with the standard control morpholino (b) or gene-specific morpholino (c–h) and monitored during development. No gross lethality or developmental abnormalities were observed at 72 h.p.f. in gene-specific morpholino-injected embryos (c–h) compared with the control (b). a and middle and right panels of b–h, lateral view, anterior left; left panels of b–e, g, h, ventral view, anterior up; left panel of f, dorsal view, anterior up. The genes appear to be nonspecifically expressed during embryogenesis as shown by patterns deposited in the ZFIN resource (http://zfin.org).