Nature - USA (2019-07-18)

Article reSeArcH

ab

# cells = 18,722 Patient ID

ET01

ET02

ET03

ET04

ET05

Relative expression

Low

CD79A

DNTT

IRF7

ELANEAZU1

MPO

SPINK2CD52HOPX

AVP

CA1

PLEK

DAD1

ITGA2B

HDC

CLC

TK1

BLVRB

PDLIM1

HBB

FCER1A

TYMS

LYZ

IRF8

E/B/M MkP EP2 EP1EP-ccMEP-ccMEPHSPC1HSPC2HSPC3IMP1IMP2IMP-ccNP1NP2M/D1M/D2PreB1PreB2

c

d

WT (n = 9,338)

MUT (n = 7,276)

NA (n = 2,108)

HSPC

IMP

NP

MEP

EP

E/B/M

M/D

PreB

MkP

# cells = 18,722

e

High

AVP HOPX CLC IRF8 AZU1

ITGA2B CA1 BLVRB PLEK DNTT

Relative expression

Low

High

t-SNE2

t-SNE1

t-SNE2

t-SNE1

t-SNE2

t-SNE1

t-SNE2

t-SNE1

t-SNE2

t-SNE1

t-SNE2

t-SNE1

t-SNE2

t-SNE1

t-SNE2

t-SNE1

t-SNE2

t-SNE1

t-SNE2

t-SNE1

t-SNE2

t-SNE1

t-SNE2

t-SNE1

t-SNE2

t-SNE1

Extended Data Fig. 4 | Integration of samples from patients with
essential thrombocythaemia and assignment of progenitor subsets.
a, t-SNE projection of CD34+ progenitor cells from samples ET01–ET05,
after integration and batch correction using the Seurat package (Methods).
b, Heat map of top ten differentially expressed genes for clusters; lineage-
specific genes from a previous publication^26 are highlighted (Methods).
c, Representative lineage-specific genes projected onto the t-SNE
representation of CD34+ cells from samples from patients with essential

thrombocythaemia. d, t-SNE projection of CD34+ cells from samples

ET01–ET05 after applying a deep generative modelling approach for

the single-cell analysis using the scVI package (Methods)^19 , showing

assignments of progenitor subsets as determined after clustering the cells

using the Seurat package. e, Genotyping data from GoT are projected onto

the t-SNE representation generated after the scVI analysis of progenitor

cells from samples ET01–ET05. Cells without any GoT data are labelled

NA (not assignable).