Science - USA (2022-04-08)

Yazaret al.,Science 376 , eabf3041 (2022) 8 April 2022 2 of 14

UMAP 1

UMAP 2

CD14

0.23

CD4

0.05

CD8a

0.07

SLC4A10

0.4

CCR7

0.05

S100B

0.08

MS4A1

0.12

KLRB1

0.05

XCL1

0.28 0

G H

-10

0

10

-10 -5 0 5

UMAP 1

UMAP 2

BMem

BIN

Plasma

DC

MonoNC MonoC

NKR NK

CD8ET

CD8S100B

CD8NC

CD4NC

CD4ET CD4SOX4

E

60

40

20

0

Pe

rcent (

%)

B

Me

m

B

IN

Plasma MonoDC

NC

Mono

C

NK

R

CD8 NK

ET

CD8

S^1

00

B

CD8

NC

CD4

NC

CD4

ET

CD4

SOX4

6 5 4 3 2 1 0

Cells (Log

) 10

F

D

PMBC

Myeloid

Monocyte

Lymphoid Sub-Lymphoid

B Cells

T Cells

NK

CD4

CD8

BIN

TCL1A

FCER2

IL4R

BMem

TACI

CD27

Plasma

IgJ

BCMA

NK

GZMA

GZMB

XCL1

XCL2

GZMK

NKR

CD8NC

LTB

CCR7

PASK

CD8ET

GNLY

NKG7

CD4NC

CCR7

SELL

LRRN3

CD4SOX4

SOX4

ID2

SELL

CD4ET

KLRB1

GZMK

TNFSF13B

CCR7 low

SELL low

MonoC

CD14

LYZ

MonoNC

CD16

DC

CST3

FCER1A

SERPINF1

Clustering

Supervised

CD8S100B

LTB

IL7R

KLRB1

GZMK

S100B

C

Cells per individual

Individuals

(^0100020003000)
50
100
150
B
−10
0
10
−10 0 10 20
UMAP 1
UMAP 2
AFR
AMR
EAS
EUR
SAS
OneK1K
A Cell-specific
eQTL mapping
Blood
from 982
individuals
genomic DNA
~1.27million
PBMCs
mRNA capture
and sequencing
SNP genotyping
Fig. 1. Population-scale scRNA-seq identifies 14 transcriptionally distinct
mononuclear populations in peripheral blood.(A) scRNA-seq data from PMBCs
were generated using a pooled multiplexing strategy for 982 healthy individuals.
Simultaneously, SNPs were genotyped, and data were integrated for single-cell eQTL
analysis. (B) UMAP analysis shows the genetic relationship between the individuals
from the OneK1K cohort and the 1000 Genomes Project ( 83 ). Individuals from
the OneK1K cohort are embedded with individuals with Northern European genetic
ancestry. AFR, African; AMR, ad-mixed American; EAS, East Asian; EUR, European;
SAS, South Asian. (C) Mean of 1291 individual cells per donor, ranging from 62 to
3501 after scRNA-seq, demultiplexing, and quality-control filtering. (D) Hierarchical
classification of cells. Each cell underwent up to four rounds of supervised clustering
based on similarity to each node, as indicated by the black arrows. After that,
unsupervised clustering by Seurat (gray arrows) yielded 14 transcriptionally distinct
cell types. Classification of each cell was confirmed based on cosine similarity to
FACS reference data and further assessed through the interrogation of differentially
expressed and prototypical genes. (E) Total percentage of each cell type as a
proportion of the total sequenced population for each individual. (F) The total number
of cells per cell type after sequencing, demultiplexing, and quality-control filtering.
(G) UMAP of 1,267,758 PBMCs across all individuals, with 14 transcriptionally distinct
populations. Color coding is the same as in (D). (H) Density plots of nine differentially
expressed canonical markers of peripheral immune cells, demonstrating robust
concordance with canonical markers (see fig S11 for additional markers). Values
denote maximal density. The abbreviations for each cell type are displayed in table S1.
The color scale is relative, ranging from no density (0) to highest density.
RESEARCH | RESEARCH ARTICLE