Evolution of Human Brain Left–Right Asymmetry: Old Genes

with New Functions

Jianguo Wang ,* ,† Sidi Ma,† Peijie Yu, and Xionglei He *

State Key Laboratory of Biocontrol, School of Life Sciences, Sun Yat-sen University, Guangzhou, Guangdong Province, China
†
These authors contributed equally to this work.
*Corresponding authors: E-mails: [email protected]; [email protected].
Associate editor: Katja Nowick

Abstract
The human brain is generally anatomically symmetrical, boasting mirror-like brain regions in the left and right hemi­
spheres. Despite this symmetry, fine-scale structural asymmetries are prevalent and are believed to be responsible for
distinct functional divisions within the brain. Prior studies propose that these asymmetric structures are predominant­
ly primate specific or even unique to humans, suggesting that the genes contributing to the structural asymmetry of the
human brain might have evolved recently. In our study, we identified approximately 1,500 traits associated with human
brain asymmetry by collecting paired brain magnetic resonance imaging features from the UK Biobank. Each trait is
measured in a specific region of one hemisphere and mirrored in the corresponding region of the other hemisphere.
Conducting genome-wide association studies on these traits, we identified over 1,000 quantitative trait loci. Around
these index single nucleotide polymorphisms, we found approximately 200 genes that are enriched in brain-related
Gene Ontology terms and are predominantly upregulated in brain tissues. Interestingly, most of these genes are evo­
lutionarily old, originating just prior to the emergence of Bilateria (bilaterally symmetrical animals) and Euteleostomi
(bony vertebrates with a brain), at a significantly higher ratio than expected. Further analyses of these genes reveal a
brain-specific upregulation in humans relative to other mammalian species. This suggests that the structural asym­
metry of the human brain has been shaped by evolutionarily ancient genes that have assumed new functions over time.
Key words: brain asymmetry, human brain, gene age, left–right asymmetry, brain evolution, brain gene expression.

Introduction brain asymmetry features are either primate-specific or

human-specific (Leroy et al. 2015; Marie et al. 2018; Graic
While the human brain largely displays bilateral symmetry, et al. 2020; Neubauer et al. 2020; Xiang et al. 2020;
with one-to-one corresponding regions in the left and Gonzalez et al. 2022; Wan et al. 2022; Hill et al. 2023).
right hemispheres, it also showcases notable structural

Article
For instance, the human brain presents a similar but
asymmetry within these corresponding regions. This asym­ more variable spatial asymmetry pattern compared with
metry underpins functional dominance in one hemi­ great apes (Neubauer et al. 2020), and there are substantial
sphere, a phenomenon referred to as brain lateralization asymmetry differences between human and chimpanzee
or hemispheric specialization (Duboc et al. 2015; Kong in terms of surface area and cortical thickness (Xiang
et al. 2018; Sha, Schijven, et al. 2021; Saltoun et al. 2023; et al. 2020). Furthermore, certain brain regions and struc­
Williams et al. 2023). Brain lateralization plays a crucial tures demonstrate unique asymmetries rarely seen in
role in various advanced neural functions, such as con­ other species (Leroy et al. 2015; Hill et al. 2023). Some brain
sciousness (Hartwigsen et al. 2021), speech (Poeppel and asymmetry patterns, however, are shared with other pri­
Assaneo 2020), language (Eckert et al. 2022), memory mates (Marie et al. 2018; Gonzalez et al. 2022). Mouse
(Bein et al. 2020), cognitive performance (Wu et al. brains, when compared, display asymmetry primarily in
2022), and handedness (Sha, Pepe, et al. 2021), among hippocampus size, with shape asymmetry observed in
others. For instance, unique thickness asymmetries in the most examined regions, but insignificant volume asym­
postcentral gyrus and inferior occipital cortex have been metry (Spring et al. 2010; Parnell et al. 2013;
associated with left-handedness (Sha, Pepe, et al. 2021). Barbeito-Andres et al. 2016). Hence, most, if not all, ob­
Additionally, abnormalities in brain asymmetry have been served brain asymmetries in human brain appear to be
linked to various neuropsychiatric disorders (Cardinale evolved recently.
et al. 2013; Li et al. 2020; Sha, Schijven and Francks 2021; In this research, we explore the evolutionary origins of
Park et al. 2022; Schijven et al. 2023). the genes implicated in human brain asymmetry.
An intriguing facet is that brain asymmetry in humans Aspects of human brain evolution, such as cortical expan­
appears to be of recent origin, given that many human sion (Won et al. 2019), are primarily driven by changes in
© The Author(s) 2023. Published by Oxford University Press on behalf of Society for Molecular Biology and Evolution.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/
licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly
cited. Open Access
Mol. Biol. Evol. 40(9):msad181 https://doi.org/10.1093/molbev/msad181 Advance Access publication August 10, 2023 1
Wang et al. · https://doi.org/10.1093/molbev/msad181 MBE
gene regulation, with additional contributions from process, where one trait is measured in a left-brain region
recently evolved genes like SRGAP2C, a product of human- and the corresponding trait is measured in the corre­
specific gene duplication that notably enhances cortico­ sponding right-brain region (fig. 1a). These traits encom­
cortical connectivity (Vanderhaeghen and Polleux 2023). pass around 100 brain regions, such as the accumbens
The pivotal question we aim to resolve is whether the and acoustic radiation, and encompass multiple features.
genes underpinning human brain asymmetry are recent These features consist of 72 pairs related to volumes of
developments or if they are ancient genes that have ac­ white and gray matter, as well as 297 pairs pertaining to
quired new functions over time. To address this, we scru­ white matter structure ascertained through diffusion
tinized variations in brain asymmetry features in MRI. Moreover, for each pair of traits, we considered two
approximately 30,000 individuals from the UK Biobank distinct time periods (see Materials and Methods).
(Bycroft et al. 2018), intending to unveil the underlying For each pair of traits (L and R), we defined asymmetry
genes. Earlier studies have begun to illuminate the genetic traits corresponding to the trait 􏽰
√����������� pair by arcsin((L −
�����������
foundations of specific brain asymmetry features. For in­ R)/ 2(L2 + R2 )) or arcsin((Ls − Rs )/ 2(L2s + R2s )) as well
stance, one research group identified 27 lead single nucleo­ as their absolute form, where Ls and Rs represent the stan­
tide polymorphisms (SNPs) after examining 73 traits linked dardized trait values of L and R, respectively (fig. 1b).
to brain asymmetry, focusing particularly on cortical thick­ Consequently, we procured 1,504 (376 × 4) brain left–
ness, surface area, and subcortical volume (Sha, Schijven, right asymmetry traits measured at two time periods
et al. 2021). Another study investigated asymmetry in (supplementary table S1, Supplementary Material online).
brain torque components, discovering 86 lead SNPs The analysis pipeline is displayed in figure 1c. The trad­
(Zhao et al. 2022). Nevertheless, these studies predomin­ itional formula ((L − R)/(L + R)) (Sha, Schijven, et al.
antly targeted specific types of brain asymmetry features. 2021) excels with exclusively positive trait values but strug­
Our study adopted a comprehensive approach. We gles when values span from positive to negative, failing to
broadened the scope to encompass a diverse range of consider their distribution. In response, we have refined
1,504 brain asymmetry traits, including both white and this formula to deepen our understanding of brain asym­
gray matter, and various magnetic resonance imaging metry and improve the detection of underlying genes.
(MRI) features across around 100 brain regions. This en­ Our updated definitions match the traditional formula
abled a more holistic perspective of brain structural asym­ for positive L and R, and normalized derived asymmetry
metry. As a result, we pinpointed 1,195 quantitative trait traits, but they uniquely consider trait value distribution.
loci (QTLs) accounting for 337 of the 1,504 brain asym­ They also offer a clearer geometric interpretation, specific­
metry traits. We also located 216 genes in proximity to ally the deviation angle from the line L = R. A thorough
the index SNPs. Functional analysis revealed significant en­ mathematical review of the advantages of our revised defi­
richment of these brain asymmetry genes in brain-related nitions over the traditional formula and their connections is
Gene Ontology (GO) terms. Furthermore, these genes ex­ detailed in the supplementary note, Supplementary
hibited significant enrichments in upregulated differential­ Material online.
ly expressed genes (DEGs) in brain tissues. To investigate
the evolutionary origins of these genes, we performed a Identification of Genetic Variants Associated with
macroevolutionary analysis based on clade-specific gene Brain Left–Right Asymmetry Traits
origins (Capra et al. 2012; Shao et al. 2019). Intriguingly, In our initial investigation, we conducted genome-wide as­
these genes appear to have significantly emerged just prior sociation studies (GWAS) for each of the 1,504 brain asym­
to the advent of Bilateria (bilaterally symmetrical animals) metry traits in the White British population during the first
and Euteleostomi (bony vertebrates with a brain), both an­ time period (phase 1). Each trait was normalized, SNPs
cient clades. These genes also showed marked upregula­ were filtered to include 8,895,704, and GWAS was imple­
tion in the human brain compared with other species. In mented through GCTA (Jiang et al. 2019). We estimated
conclusion, our findings suggest that genes contributing the genetic relatedness matrix (GRM) using SNPs overlap­
to brain asymmetry originated in ancient times but have ping with linkage disequilibrium (LD)-pruned HapMap3
since evolved to perform new functions. SNPs and considered covariates including age, sex, age2,
age × sex, and age2 × sex as well as the top 20 principal
Results components, as outlined in a previous study (Jiang et al.
2019). Summary statistics for all 1,504 asymmetry traits
Definition of Left–Right Brain Asymmetry Traits were obtained, and at a significance level of P < 5 × 10−8,
We gathered human brain traits captured via MRI from we discerned significant genetic variants, defining the
the UK Biobank (Bycroft et al. 2018). These traits were col­ QTLs via clumping analysis (Purcell et al. 2007) for each
lected from approximately 30,000 genotyped British indivi­ trait. Consequently, 1,195 QTLs were identified for
duals representing various ancestries. Among the collected 337 of the 1,504 traits (fig. 2a; supplementary table
traits, 376 pairs correspond to measurements of symmet­ S2, Supplementary Material online; see Materials and
rical brain regions. The set of 376 trait pairs includes all Methods). The effect sizes, represented by the β coefficients
available brain traits obtained during the download of these QTLs, are displayed along with their respective P

2
Evolution of Human Brain Left–Right Asymmetry · https://doi.org/10.1093/molbev/msad181 MBE
(a) (b)
A L R
plan- : arcsin( )

8
1
2( L2 R2 )

Right (R )
L R
plan- 1 : arcsin( )
M 2
2( L2 R2 )

4
T

Ls Rs
1 plan- 2 : arcsin( )
2
2( L s R s2 )

0
O Ls Rs
0 4 8 plan- 2 : arcsin( )
Left and Right Left (L ) 2( L2s R s2 )
FIG. 1. Defining brain asym­ ACR regions
metry traits. (a) An illustration
of a pair of traits in a left brain (c)
region and a corresponding GWAS in phase 2 Functional analysis
right brain region, with ACR re­ (non-White British in UK Biobank, GO, GTEx
presenting anterior corona ra­ the first time period
diata. (b) The portrayal of 337 BA traits with QTLs called in phase 1)
brain asymmetry traits accord­ Gene anotation
ing to four different plans. (PhenoScanner, ANNOVAR)
Point M represents the popula­ GWAS in phase 1
tion averages for trait pair (White British in UK Biobank,
(L and R). Point T represents the first time period,
observed trait values for an in­ Macroevolution analysis
1504 Brain Asymmetry (BA) traits)
dividual’s trait pair. Point O de­ (GenTree, ProteinHistorian)
notes the origin. The dashed
line OA represents L = R, with
GWAS in phase 3
Ls and Rs being the standar­
(White British in UK Biobank, Evolution of gene expression
dized values of L and R, respect­
ively. (c) The overall analytical the second time period, (Six species, six tissues,
framework of this study. 337 BA traits with QTLs called in phase 1) 14 stages in life span)

values in supplementary figure S1, Supplementary Material 1 and 3 (SCR = 0.69, P < 8.65 × 10−90, binomial test). We
online. For instance, the asymmetry trait of plan-I, corre­ calculated Pearson’s R between the β coefficients of top
sponding to weighted-mean MO (diffusion tensor mode) significant SNPs in phases 1 and 2 or 3, finding a significant
in the superior thalamic radiation tract, contained 12 correlation for most traits between phases 1 and 3 (fig. 2d
QTLs. The Q–Q plot for this trait demonstrated effective and e; supplementary fig. S2, Supplementary Material on­
control of the population structure (inflation factor line), but not between phases 1 and 2 due to the smaller
λ = 1.018; fig. 2b), and the Manhattan plot exhibited the sample size and complex population structure in phase 2
genome-wide distribution of significant genetic variants (see Materials and Methods) (Makowski et al. 2022).
(fig. 2c). As the original brain MRI traits derive from iden­ These results from phases 2 and 3 underscore the robust­
tical MRI images for the same subjects, the population ness of the identified genetic variants against population
structure should be well controlled for other traits. structure and temporal variation.
We then performed a replication analysis in British indi­
viduals of other ancestries (phase 2) (see Materials and Gene Annotation and Functional Enrichment
Methods), procuring summary statistics for each of the We began by associating the genes with each of the 1,195 in­
337 traits and testing the significance of QTLs identified dex SNPs using two software tools: PhenoScanner (Kamat
after clumping in phase 1 at P < 0.05/n = 4.56 × 10−5, et al. 2019) and ANNOVAR (Wang et al. 2010) (see
where n = 1,096 represents the count of QTLs shared in Materials and Methods). Noting a high degree of overlap in
phase 2 (Makowski et al. 2022). A total of 19 QTLs re­ the genes annotated by these two tools (supplementary fig.
mained significant in phase 2, a significant outcome based S3, Supplementary Material online), we combined the anno­
on a binomial test (P = 1.28 × 10−42). The sign concord­ tations from both, resulting in 216 genes (supplementary
ance rate (SCR) of top significant SNPs between phases 1 table S3, Supplementary Material online). These are hence­
and 2 was also statistically significant (SCR = 0.54, forth referred to as “brain asymmetry” (BA) genes. Notably,
P = 8.65 × 10−90, binomial test) (see Materials and some of the BA genes we identified, such as MAP2 and
Methods) (Makowski et al. 2022). MAPT known for their involvement in brain asymmetry
In the subsequent replication analysis in the White and neurodegenerative diseases (Lubben et al. 2021), have
British population measured at a second time period also been implicated in previous studies (supplementary
(phase 3) (see Materials and Methods), the number of table S3, Supplementary Material online).
QTLs remaining significant was also noteworthy (n = 1, Subsequently, we performed a GO analysis on these
P = 0.049, binomial test), as was the SCR between phases genes (see Materials and Methods) (Wu et al. 2021),
3
Wang et al. · https://doi.org/10.1093/molbev/msad181 MBE

FIG. 2. GWAS of brain asym­

metry traits. (a) The distribu­
tion of QTL numbers for the
337 identified traits each hav­
ing at least one QTL. The effect
sizes and corresponding P va­
lues are depicted in
supplementary figure S1,
Supplementary Material on­
line. (b) The Q–Q plot illus­
trates the well-controlled
population structure for an ex­
ample asymmetry trait defined
by plan-I for the weighted-
mean MO in the tract superior
thalamic radiation of left and
right hemispheres. MO stands
for diffusion tensor mode. Red
and gray colors represent ex­
pected and observed P values,
respectively. The inflation fac­
tor λ = 1.018 is very close to
1. (c) The Manhattan plot for
the trait used in (b), with
ten QTLs positioned within
the range Ch17: 43460181–
44821987. The red dashed line
marks the significance thresh­
old P < 5 ×10−8, and the index
SNPs corresponding to QTLs
are highlighted in yellow. (d )
The scatter plot illustrates a
significant positive Pearson’s
correlation (R) between the β
coefficients of phase 1 and
phase 3 for the trait examined
in (b). (e) The Pearson correl­
ation (R) between the β coeffi­
cients of phase 1 and phase 3
for all of the 337 traits. The cor­
responding P values are de­
picted in supplementary
figure S2, Supplementary
Material online.

finding significant enrichments in brain-related biological We used GTEx gene expression data (GTEx
processes, such as axonogenesis, and cellular components Consortium 2020), accessed via FUMA online, to com­
like the synaptic membrane (fig. 3a). We also found enrich­ pare the expression values of each identified gene be­
ments related to the microtubule structure, essential for tween brain and nonbrain tissues. Based on their
neural axons and dendrites, consistently present across significance, the identified genes were categorized
biological processes, molecular functions, and cellular into three classes: significantly upregulated, signifi­
components (supplementary table S4, Supplementary cantly downregulated, or insignificant in brain-related
Material online). This suggests the functional relevance tissues when compared with nonbrain tissues (fig. 3c;
of our identified genes. supplementary table S3, Supplementary Material on­
We then utilized FUMA (Watanabe et al. 2017) to assess line; see Materials and Methods). Roughly a third of
whether BA genes were enriched in upregulated or down­ the identified genes, significantly upregulated or
regulated DEGs in both brain and nonbrain tissues (see downregulated in brain tissues, could potentially con­
Materials and Methods). Our findings revealed that BA tribute to the formation of brain-specific asymmetry
genes were significantly enriched in upregulated DEGs in features. In contrast, the remaining genes may have
brain tissues and were generally enriched in downregu­ universal roles in asymmetry across both brain and
lated DEGs in nonbrain tissues (fig. 3b). nonbrain tissues.
4
 (c) The gene expression levels of BA genes in brain and nonbrain tissues are presented (from online FUMA using GTEx gene expression data).
FIG. 3. Functional enrichments of BA genes. (a) The displayed GO enrichments of BA genes involve biological processes and cellular components,
derived using the R package “clusterProfiler.” The top 10 GO terms are presented, with the adjusted P value obtained through the hypergeo­
metric test and adjusted by the Benjamini–Hochberg method. (b) The enrichments of BA genes in upregulated and downregulated DEGs of

5
brain and nonbrain tissues are shown. The P value is obtained via a hypergeometric test. Red bars denote P < 0.05 post-Bonferroni correction.

Each row represents one of the 200 BA genes, and each column denotes a tissue. The expression level in each cell is the average log2(Transcripts
Per Million [TPM]) among replicates (as reported by online FUMA). For each gene, it is determined whether the gene expression levels in brain
tissues are significantly upregulated or downregulated compared to nonbrain tissues using a t-test after Bonferroni correction (P < 0.05/n, where
MBE

log2TPM
Cellular component Adjusted P

0
Up-regulated Down-regulated
od od
0.04
0.01

lo lo
Count

_B _B
12
15

le le
ho
9

W na ho
gi um c W ina um c
le g
Va rus _I g bi Va rus le
_I g bi
te d al le u al le u
Proportion of genes

U oi in er_ rap te d
PBon<0.05

yr U roi in er_ rap

r m w up y r m w up
Th tis h Te Lo _S Th tis h Te Lo _S
s c
Te ma e_ d_ d s c e_ d_ d
0.08

in e e Te a in e e
o st os os om st os os
St leen nte xp xp St leen nte xp xp
I E E I E E
Sp all_ un_ un_ Sp all_ un_ un_
S
Smn_S ot_ Smn_S ot_
S
i i
Sk n_Ne Sk n_Ne
i t i t
Sk ta Sk sta
os r y o ry
Pr ita s Pr ita s
tu a tu a
Pi re Pi cre
Evolution of Human Brain Left–Right Asymmetry · https://doi.org/10.1093/molbev/msad181

nc l d

Up-regulated
ia al n n al al nd
Pa ry ib let la
0.05

i
Pa ry ib let la
va _T e _G va _T e _G
O rve _Sk ary O rve _Sk ary
syna ical par t omain
adin brane

bra membra brane

od e
tubule
microg edge
c c mem cell
ap microd raft
em ium
lam embran x

e v
N scle ali e
N scle ali
v
or te

u S u S
M or_ M or_
e
of

cell c
n

in l a in lla
ip

l Non-brain tissues
Non-brain tissues

M
ng du M
ng du
ll

e e
Lu r Me ex icl ag e e
ma m
ptic m e

tio
n Lu r Me ex icl ag n
ve _ rt tr d ve _ rt tr d tio
Li ney Co Ven pen
le

nc Li ey o en en nc
pti

Ju C
d _ _ p dn _ _V pp Ju
e
ll

Ki ney eft l_A ris l_ Ki ney eft l_A

l plas
e

ris l_
n
a

d L la ea ea
n

Ki art_ tria be cu d L la
ag Ki art_ tria be cu ag
y

e A Tu s a h
posts

H rt_ n_ Mu os op e A Tu s a h
mem

n is the number of genes). Notably, 16 genes are not included in the FUMA results.
H rt_ n_ Mu os op
apica

ea ia _ c es ea ia _ c es
H p us u ro H lop gus Mu tro

Insignificant
llo ag _M st l a _ s
Fa ph gus Ga e es Fa ph gus Ga e es
yt
Biological process Adjusted P

o a _ rs o a _ rs yt
Es ph gus sve oc Es ph gus sve oc
0.03
0.01

o a ph o a ph
Count

Es ph Tran oid vix m Es h an id ix m

ly op Tr mo rv
13
18

o m r ly
Es lon_Sig oce ix ed_ ts Es lon_Sig oce ix ed_ ts
8

o d v s d v
C on_ En cer orm la .1
o
C lon_ En cer formbla e .1
s
ol _ to sf ob e c o _ to s o c
C vix Ec an ibr ssu al_ C vix Ec an ibr ssu al_
of genes

er _ .tr _f Ti ic er _ .tr _f Ti ic
C rvix BV red ry_ erv C vix BV ed y_ erv
e E a c er E r ar c
C lls_ ultu m rd_ C lls_ ultu m rd_
0.10

e C m o e C m o
C lls_ Ma l_c C lls_ Ma l_c

Down-regulated
e _ a e _ a
C st in C st in
ea Sp ea Sp
Br n_ Br in_
ai r ial ry a r ial ry
Proportion

Br de ib na tu
m Br de ib na um
ad _T ro en lia ad _T ro t
lia
Bl ry o ng Bl ery Co a en
te _C ta m m ng
Ar ry or d _O us ga t _
Ar ry or d _O us
t
ga
te _A lan ral o lia al_ te _A lan ral o lia al_
Ar ery l_G ce tanegra ang as Ar ery l_G ce tanegra ang as
t a is i g b
0.05

t a is i g b
Ar en _V bcu a_n al_ ns_
r Ar ren _V bcu a_n al_ ns_
e u ti s e e u ti s e
Adipos _S tan ba mb Adipos _S tan ba mb
e s _ u e s _ u
Adipos ub en acc Adipos ub en acc
org ent size

rane
in loction to pulation oftranspor t
reg cellular nization
synar componuidance
on o uron pro axon geriphery
axonanization

n to a memb size
ula ction g idance
esis

S m _ s S m _ s
Ad in_ uta us mu Ad n_ ta s u
A9 ai Pu leu lam s A
9
ogen

a P e a l l s
cell

Br n_ c a u _B Br _ c a u B
ai Nu oth mp ex n
ai Nu oth mp ex
_
u

Br n_ p a rt
ra n orga

Br in_ yp ca ort
cell p

ai H y oc C o re 4

Brain Vs non-brain :
Brain tissues
a H o C re 4
Brain tissues

Br in_ ipp al_ he A2 Br in_ ipp al_ he A2

a H t sp lia _B a H t sp lia _B
lasm

Br in_ ron x m mi ng tex

e

Br in_ ron x m mi ng tex

tio
s

je

a F e u e a r a F e u e a r
p

Br n_ rt ell _H _g co Br in_ ort ell r_H l_g _co

rojec

ai Co b ar al e_ a C b a
o

Br n_ re ell as at Br n_ re ell as at
a e
ti

ai Ce b _b ul ai Ce b _b ul
f cell

Br in_ ere ate ing Br in_ ere ate ing

p
aliz

n
a C d c
Br in_ au ior_ a C d c ai
ll

Br n_ u r_
e

br
ne

a C r la ai Ca rio la
p caliza

fc

Br in_ nte da Br n_ te a
ai An gd n-
o

a A g
Br in_ my Br in_ my
ti

no
n
te
la

a A
o
in lo

Br in_ a A
Br in_
u

ro

cytose regulati
Vs
reg

a a
Br Br
prote
in
10 5 0 ra
k
8 4 0
B

tiv
log10(P) log10(P)

nega

(b)

(c)
(a)

Brain-asymmetry Genes (n=216)

Wang et al. · https://doi.org/10.1093/molbev/msad181 MBE
Evolutionary Origin of BA Genes is observable among Euteleostomi species (Hibi and
In a prior study, GenTree (Shao et al. 2019) mapped the Shimizu 2012; Briscoe and Ragsdale 2019; Hain et al.
origins of human protein-coding genes onto a phylogenet­ 2022). It is therefore feasible that the genes responsible
ic tree via synteny comparison, delineating the clade- for brain formation also contribute to brain asymmetry.
specific origins of genes. For each clade, we calculated Collectively, our findings propose that BA genes originated
the proportion of BA genes and compared these propor­ from two significant evolutionary transitions: the emer­
tions with those of the total background genes. Using a bi­ gence of bilaterally symmetrical animals and the formation
nomial test for each clade, we found that the ancient clade, of the brain.
Euteleostomi, demonstrated significant enrichment for BA In addition, we considered several potential factors that
genes (fig. 4a; supplementary table S5, Supplementary might influence the results. First, we amalgamated the an­
Material online; see Materials and Methods). Since cient clades preceding Tetrapoda in the ProteinHistorian
GenTree relies on synteny, which is accurate for recently tree into an “ancient” clade and reperformed the enrich­
originated clades but loses resolution for more ancient ment analysis on this degenerated tree. The findings
clades, our findings suggest that genes underlying brain showed consistent enrichment between the simplified
asymmetry are not newly evolved. ProteinHistorian tree and GenTree (supplementary fig.
To further illustrate the evolutionary origins of BA genes, S4, Supplementary Material online). Second, we compared
we used another set of clade-specific genes as a background, the significance of each clade between BA genes defined by
constructed based on protein family information from an­ plan-I or plan-III and those defined by plan-II or plan-IV, re­
other study (ProteinHistorian; Capra et al. 2012). vealing significantly positive correlations between nonab­
ProteinHistorian offers improved resolution in ancient clades, solute and absolute definitions of brain asymmetry
albeit with some accuracy compromise compared with (supplementary fig. S5, Supplementary Material online).
Third, we analyzed clade enrichments solely for significant­
GenTree. Our findings indicated that BA genes are significant­
ly upregulated and downregulated genes in brain tissues as
ly enriched in two ancient clades, Bilateria and Euteleostomi
defined in figure 3c, which demonstrated similar enrichments
(fig. 4b; supplementary table S5, Supplementary Material on­
in ancient clades (supplementary fig. S6, Supplementary
line; see Materials and Methods).
Material online; see Materials and Methods). Lastly, we con­
Bilateria is associated with the origin of bilaterally sym­
sidered the nonindependence between statistical tests for dif­
metrical animals (Finnerty et al. 2004). Left–right asym­
ferent clades and implemented a stepwise statistical test
metry only holds significance within the context of
procedure. The results still showed ancient enrichments for
bilateral symmetry. Prior research indicates that bilateral
both trees (supplementary fig. S7, Supplementary Material
asymmetry may offer certain advantages, though not as
online; see Materials and Methods).
universally as bilateral symmetry (Genikhovich and In conclusion, the majority of BA genes originate from
Technau 2017; Vallortigara and Rogers 2020). It is plausible ancient clades, specifically Bilateria and Euteleostomi,
that this trade-off between bilateral symmetry and bilat­ where they are significantly enriched, while a minor portion
eral asymmetry originated in early Bilateria and has been are of more recent origin as detailed in supplementary table
inherited by extant bilateral organisms. S5, Supplementary Material online.
Euteleostomi, also known as bony vertebrates, is the com­
mon ancestral clade of Sarcopterygii and Actinopterygii, the
former including all tetrapod species and the latter most fish Evolution of Gene Expression Underlying Human
species (reference: the life map in NCBI, https://lifemap-ncbi. Brain Asymmetry
univ-lyon1.fr). The Euteleostomi clade is renowned for its Considering the relatively recent evolution of most brain
evolution of a complex brain structure—comprising the asymmetry traits, we are particularly interested in under­
forebrain, midbrain, and hindbrain—hereafter simply re­ standing how ancient genes shape these asymmetric char­
ferred to as the “brain” (Šestak and Domazet-Loso 2015). acteristics in the human brain. We propose that these
This complexity stands in contrast with the more rudi­ genes have gradually developed new functions over time.
mentary brain structures, such as the nerve cord or neural To investigate this hypothesis, we compiled gene expres­
tube, seen in the primitive members of the Chordata clade. sion profiles from six organs (brain, liver, heart, kidney,
The transition from a simpler to a more complex brain ovary, and testis) across six species (human, macaque,
structure during the evolution from Chordata to mouse, rat, rabbit, and opossum) as reported in a prior
Euteleostomi coincides with the emergence of three sig­ study (Cardoso-Moreira et al. 2019). This research assessed
nificant anatomical features: the spine, jaw, and bones. gene expressions in these organs at 14 distinct develop­
The spine provides structural support, the jaw enables a mental stages, which were harmonized and made compar­
wider range of feeding mechanisms and diets, and the able across species by integrating developmental
bones offer enhanced bodily integrity and protection. transcriptome data and referring to the 23 Carnegie stages
These developments, in turn, may have created the condi­ that segment embryonic development (Cardoso-Moreira
tions for the evolution of the complex brain structure, in­ et al. 2019). The corresponding stage relationships are de­
dicating a nuanced interplay of selective pressures. As picted in figure 5a. We noted that certain stages in one
earlier studies suggest, a common basic brain architecture species may correspond to multiple stages in others; for
6
Evolution of Human Brain Left–Right Asymmetry · https://doi.org/10.1093/molbev/msad181 MBE
(a)
BA genes Total genes
(n=176) (n=19462)
Homo Sapiens Homo sapiens
Homininae
Hominidae Pan troglodytes
Hominoidea Pongo pygmaeus abelii
Catarrhini Nomascus leucogenys
Simiiformes Macaca mulatta
Euarchontoglires Callithrix jacchus
Boreoeutheria Mus musculus,Oryctolagus cuniculus,etc.
Eutheria Canis lupus familiaris,Bos taurus,etc.
Theria Loxodonta africana
Mammalia Monodelphis domestica
Amniota Ornithorhynchus anatinus
Tetrapoda Gallus gallus,zebra finch,etc.
P=8.8×10-9 Euteleostomi Xenopus tropicalis
Danio rerio,Tetraodon nigroviridis,etc.

0.8 0.4 0.0 0.4 0.8 GenTree

Proportion of genes

(b) BA genes Total genes

(n=169) (n=17936)

Bony vertebrates with a brain

Bilaterally symmetrical animals

HUMAN Homo sapiens
Homininae
Catarrhini Pan troglodytes
Euarchontoglires Rhesus
Eutheria Murinae
Theria Laurasiatheria
Mammalia Monodelphis domestica
Amniota Ornithorhynchus anatinus
Tetrapoda Gallus gallus
P=0.008 Euteleostomi Xenopus tropicalis
Chordata Clupeocephala
Deuterostomia Ciona intestinalis
P=0.002 Bilateria Strongylocentrotus purpuratus
Opisthokonta Ecdysozoa
Eukaryota Ascomycota
Cellular Organisms Other Eukaryota
Bacteria and Archaea

0.3 0.2 0.1 0.0 0.1 0.2 0.3

ProteinHistorian
Proportion of genes

FIG. 4. Tracing evolutionary origins of BA genes. (a) The left bar graph contrasts the proportion of BA genes in each clade with the total genes’
proportion per clade. A few BA genes that do not overlap with total genes are not considered. The 1:1 corresponding clades, derived from the
topological life tree (GenTree, Shao et al. 2019), are presented to the right. The red bar indicates statistical significance based on the binomial test
with the corresponding P value shown. (b) Similar to (a), except that the ProteinHistorian tree (Capra et al. 2012) is considered. Bilateria and
Euteleostomi correspond to two key evolutionary transitions: the emergence of bilaterally symmetrical animals and the formation of the brain.

instance, stage 7 weeks postconception (wpc) in humans genes for further analysis, as the cerebellum was considered
matches two stages in mice, rats, and rabbits (fig. 5a). a distinct organ in the original study (Cardoso-Moreira et al.
Additionally, it is important to mention that data for 2019) (see Materials and Methods).
stages 1–9 in macaques are incomplete (fig. 5a; see We introduced the Brain Specificity Index (BSI) of a
Materials and Methods). gene, defined as the log2 of the fold change (FC) between
Orthologous information was obtained from the the gene’s expression in the brain and the average gene ex­
Ensembl database, and we retained any human gene that pression in other organs (see Materials and Methods). The
has a unique ortholog in at least one of the other five species BSI was calculated for each of the 158 BA genes across the
for further analysis. This process yielded orthologous infor­ six species and at each of the 14 developmental stages
mation for 172 out of the 216 identified genes across the six (supplementary table S6, Supplementary Material online;
species. However, we excluded 14 genes specifically asso­ see Materials and Methods). When conducting a compari­
ciated with cerebellum-related traits, thereby leaving 158 son for a specific gene between two species, we selected
7
Wang et al. · https://doi.org/10.1093/molbev/msad181 MBE
(a) Stage
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Human
4wpc 5wpc 7wpc 7wpc 8wpc 11wpc 12wpc 13wpc 19wpc 19wpc 20wpc toddler youngAdult youngMidAge

Macaque
NA NA NA NA NA NA NA NA NA e108 e130 P23 P152 P3285

Mouse
e10.5 e11.5 e12.5 e13.5 e14.5 e15.5 e16.5 e17.5 e18.5 P0 P3 P14 P28 P63

Rat
e11 e12 e14 e15 e16 e18 e19 e20 e20 P0 P3 P14 P42 P112 FIG. 5. Comparative analysis of
BA genes’ BSI across multiple
Rabbit
e12 e13 e15.5 e16.5 e18 e19.5 e21 e23 e27 e27 P0 P14 P84 P186 species. (a) The diagram
Opossum displays the corresponding rela­
13.5 14 16 16 18 18 20 28 28 28 35 74 104 134 tionships across the develop­
mental stages of different
(b) (d) species. The term “NA” denotes
Comparison of BSI between human and other species Example genes at stage 13 the unavailability of data for a
1.25×10-2 (n=87) specific stage. (b) Pairwise box­
ENSG00000073921 | PICALM
2.50×10-3 n=104) plots illustrate the comparison

0.50
0.25
5.22×10-5 (n=98)
of the BSI for BA genes between

0.00
1.35×10-6 (n=116)
humans and each of the five
BSI at stage 13 (log2FC)

7.88×10-3 n=106
other species at stage 13, corre­
5
sponding to the young adult
ENSG00000074590 | NUAK1
stage in humans. Below, the spe­
0 cies tree is represented topo­
3
logically. The P values of each
2
1
pair, calculated using a one-
tailed pairwise Wilcoxon test,
ENSG00000075399 | VPS9D1 are presented. The red dashed
0.50.70.91.11.3

line indicates a BSI of zero. The

BSI for each gene is computed
BSI of individual genes (log2FC)

opossum rabbit rat mouse macaque human as the log2-transformed FC be­

tween its brain expression and
ENSG00000076356 | PLXNA2 its average expression in other
0.00.51.01.52.0

organs (here, the brain vs. the

Species
heart, liver, and testis for each
species pair), denoted as log2­
(c) (FC). (c) This panel displays the
Comparison of BSI difference across stages
ENSG00000122786 | CALD1 comparative analysis of the
6

human_macaque overall BSI of BA genes between

human_mouse humans and each of the other
human_rat
species across all 14 develop­
log10P for BSI difference

human_rabbit
human_opossum mental stages. The P values, cal­
culated using a one-tailed
4

ENSG00000138821 | SLC39A8
pairwise Wilcoxon test, are
shown in logarithmic scale.
The abbreviation “wpc” refers
to weeks postconception. At
2

ENSG00000158805 | ZNF276 each stage, a comparison is con­

ducted between humans and
0.0

each of the other species,

wherein the BSIs for both spe­
0

cies are calculated based on

1 2 3 4 5 6 7 8 9 10 11 12 13 14
the organs they have in com­
opossum

rabbit

rat

mouse

macaque

human
c

c
pc

pc

pc

mi ung
yo r
ad ng
dle

mon. (d) This section presents

p

wp

wp

wp

wp

wp

wp

ult

ag le
4w

5w
7w

7w

8w

e
dd
u

yo
tod
11

12

13

19

19

20

examples of genes that exhibit

Stage adults Species higher BSI in humans during
stage 13.

common organs devoid of missing data and with an ex­ Methods). We assessed the robustness of the RPKM > 1
pression level satisfying the Reads Per Kilobase Million criterion by comparing it with a lower threshold (RPKM
(RPKM) > 1 criterion for the calculation of the BSI. We > 0.5) and a higher one (RPKM > 1.5), observing a repro­
then established the significance of the BSI in humans ducible pattern across these criteria (supplementary fig.
being typically larger than in any of the other species using S8, Supplementary Material online). Notably, stages 13
a one-tailed pairwise Wilcoxon test (supplementary table and 14, representing young adult and young middle age,
S6, Supplementary Material online; see Materials and respectively, closely correspond to the age range (40–69

8
Evolution of Human Brain Left–Right Asymmetry · https://doi.org/10.1093/molbev/msad181 MBE
years) in which brain traits are typically measured in the Material online) (Sha, Schijven, et al. 2021). When we ap­
UK Biobank. plied our macroevolution analysis to these 52 genes using
Initially, we focused on stage 13, which correlates with the ProteinHistorian, we found no significant enrichments in
young adult stage in humans. The overall BSI of BA genes is the two ancient clades we identified (supplementary fig.
significantly higher in humans compared with each of the S9a, Supplementary Material online). This discrepancy
other species (fig. 5b; see Materials and Methods), suggesting could be due to the limited trait sampling and gene iden­
a brain-specific upregulation in humans relative to other spe­ tification in Sha’s study compared with our study. Brain
cies. We then investigated when the brain-specific upregula­ asymmetry can be categorized into two types: torque
tion of BA genes in humans manifests during development, asymmetry, involving asymmetry in 3D space (Li et al.
from 4 weeks postconception to young middle age. Our re­ 2018), and asymmetry within corresponding brain regions.
sults indicate that the brain-specific upregulation of these We focused on the latter due to its strong association with
genes in humans is gradually established during development advanced neural functions. A previous study investigating
and becomes consistently significant in adults (fig. 5c; see torque asymmetry identified 68 genes linked to its compo­
Materials and Methods). Sporadic signals were detected at nents (supplementary table S5, Supplementary Material
4–7 weeks postconception, 13 weeks postconception, and online) (Zhao et al. 2022). When analyzing these genes
during the toddler stage when comparing humans with using ProteinHistorian, we discovered subtle yet significant
mice, rabbits, and rats (fig. 5c). Previous studies on brain enrichments in the two ancient clades identified
asymmetry features in infants have shown that certain traits (supplementary fig. S9a, Supplementary Material online).
gradually develop from infancy to adulthood, such as the Notably, the gene sets from Sha’s study and Zhao’s study
brain volumes of subcortical regions (Dean et al. 2018). overlap significantly with ours (supplementary fig. S9b,
Additionally, specific brain asymmetry traits remain consist­ Supplementary Material online) and show enrichments in
ent from infancy to adulthood, as observed in the four fea­ the same ancient clade of GenTree (supplementary fig.
tures within perisylvian regions (Glasel et al. 2011). The S9c, Supplementary Material online). By integrating these
early establishment of brain asymmetry is thought to con­ genes into our study, we found consistent enrichment re­
tribute to the development of cognitive competencies sults (supplementary fig. S9a and c, Supplementary
(Dehaene-Lambertz and Spelke 2015), suggesting a potential Material online), reinforcing that both types of brain
evolutionary pressure (Glasel et al. 2011). asymmetries share roots in the same ancient clades. Our
We present examples of genes exhibiting higher BSI in larger gene set showed more pronounced significance
humans at stage 13 (fig. 5d). Notably, at stage 13, gene ex­ (supplementary fig. S9d, Supplementary Material online).
pression profiles are available for four organs in humans: We also explored asymmetrical gene expression be­
the brain, heart, liver, and testis. These organs are also pro­ tween the left and right brain hemispheres. Our aim was
filed in the other five species, making stage 13 more suit­ to determine whether the BA genes identified in this study
able for comparison than stage 14, where the number of exhibit a higher degree of asymmetrical expression than
organs varies between different species pairs. Integrating other genes. Using RNA-Seq data of two donors derived
the results of enrichment analysis for BA genes in upregu­ from the Allen Brain Atlas (http://human.brain-map.org/
lated DEGs in brain-related tissues (fig. 3b), our compara­ static/download), we computed the asymmetry levels
tive analysis of BSI between humans and other species (ALs) for each gene across around 30 paired brain regions
collectively supports the proposition that BA genes have for each donor (supplementary table S7, Supplementary
acquired new functions over time. Material online; see Materials and Methods). We also aver­
aged these calculations from two donors for their common
brain regions considering the subtle yet statistically signifi­
Discussion cant correlation between them (R = 0.036, P < 2.2 × 10−16;
In this study, we cataloged approximately 1,500 brain see Materials and Methods). We did not observe a signifi­
asymmetry traits across nearly 100 primary brain regions, cant increase in ALs for the BA genes compared with other
representing the most comprehensive investigation into genes (P = 0.91 for donor 1, P = 1.00 for donor 2, and P =
brain asymmetry to date, to the best of our knowledge. 1.00 for their average, one-tailed Wilcoxon test).
We systematically identified the genes associated with these Additionally, we evaluated each brain region separately
traits, tracing their origins back to two crucial evolutionary and identified two regions for donor 1, one region for do­
transitions: the emergence of bilaterally symmetrical animals nor 2, and one region for their average with a P value less
and the formation of the brain. Furthermore, we evaluated than 0.05, albeit these did not pass multiple testing (see
the likelihood of these ancient genes evolving new functions Materials and Methods). Our results align with a previous
to drive brain asymmetry by analyzing their evolution in gene study that examined three cortical regions associated with
expression. handedness and language in humans and their counter­
Our research covered a wide range of brain asymmetry parts in rhesus macaques. This study found that gene ex­
features, identifying the most extensive set of BA genes to pression profiles were generally similar across both
date. In contrast, a previous study examining 73 brain hemispheres, except for the human posterior superior
asymmetry traits identified only a quarter of our total, temporal cortex that displayed a distinct left–right differ­
52 genes (supplementary table S5, Supplementary ence (Muntané et al. 2017). Another study highlighted
9
Wang et al. · https://doi.org/10.1093/molbev/msad181 MBE
differential microRNA expression between the left and valence lateralization model for emotional lateralization
right hemispheres, suggesting a potential regulatory role (Palomero-Gallagher and Amunts 2022). Among these
of microRNA in establishing brain asymmetry (Miao theories, the Geschwind–Galaburda–Behan model holds
et al. 2020). Several factors might have contributed to considerable significance. This model proposes that ele­
the lack of a significant increase in ALs for BA genes. vated fetal testosterone levels can retard left hemisphere
These include no overlap or sparse sampling in brain re­ development, potentially resulting in right-hemisphere
gions, asymmetry traits being contributed by specific cells dominance, left-handedness, immune disorders, and dys­
rather than all cells, timing mismatches between asym­ lexia (Geschwind and Galaburda 1985). Remarkably,
metry trait determination and gene expression measure­ among the BA genes we identified in our study, KANSL1
ment, or the limited availability of gene expression data stands out—a gene previously reported to underlie dys­
from only two individuals. lexia (Paracchini et al. 2016). Manns’ (2021) multilevel
The origins of brain structural asymmetry have long fas­ model postulates that genetic factors instigate basic brain
cinated scientists. Identifying conserved brain regions and asymmetry during embryonic development, which is then
structural asymmetries is a crucial step in decoding the sculpted by functional needs, subsystem interactions, and
evolutionary process. While some asymmetries are found environmental influences. This theory harmonizes with
in model organisms like mice (Barbeito-Andres et al. the gradual establishment of the brain-specific upregula­
2016) and zebrafish (Duboc et al. 2015), many are unique tion of BA genes during development. Additionally, the
to humans or primates, distinguishing them from other “from hand to mouth” theory presents an intriguing con­
species. Challenges in identifying these conserved features nection between primate gestural communication and
across various species hinder progress (Leroy et al. 2015; analogous human language regions (Corballis 2002).
Marie et al. 2018; Graic et al. 2020; Neubauer et al. 2020; Empirical cross-species studies have significantly enriched
Xiang et al. 2020; Wan et al. 2022). Uncovering the BA this field. For instance, mustached bats demonstrate a left­
genes could provide insights into the evolutionary me­ ward asymmetry for social vocalizations in the primary
chanisms underlying brain structural asymmetry, inform­ auditory cortex, similar to humans (Washington et al.
ing several crucial areas of research. For instance, BA 2021). Also, larger-brained bird species have been reported
genes could serve as markers for determining correspond­ to show stronger foot preferences, analogous to human
ing brain regions across various species. Researchers can hand preferences (Kaplan and Rogers 2021). Importantly,
also explore how BA genes acquire new functions by exam­ the BA genes we identified in our study could serve as mo­
ining the variants in their gene sequences, along with their lecular markers, providing a foundational basis for these
regulatory sequences. It would be worth investigating theoretical models and offering evolutionarily conserved
whether the asymmetrical contribution of BA genes to brain molecular evidence for empirical studies.
asymmetry during development could be spatial (asymmet­ Despite the significant insights, our study has some lim­
rical gene expression between two hemispheres), temporal itations. These include the need for further investigation
(asynchronized gene expression between two hemispheres into the evolutionary mechanisms of BA genes and the
during development) (Palomero-Gallagher and Amunts mechanisms driving their upregulation. There is also the
2022), or both. Understanding how this asymmetry devel­ challenge of exploring the interplay between BA genes
ops prematurely or belatedly in the developmental process and environmental factors impacting brain asymmetry.
and its contribution to evolutionary pressure presents a Among the BA genes we identified, several are of recent
worthy challenge. Such an understanding could prove bene­ origin (supplementary table S5, Supplementary Material
ficial for studies related to associated diseases. online) and, despite not displaying a significantly higher ra­
Symmetry in animals often signifies developmental ro­ tio than expected, they demand further study due to their
bustness, underpins the “good genes” hypothesis of sexual potential contributions to brain asymmetry. Our gene
selection (Byers and Waits 2006), and influences art and identification method could be enhanced through the
aesthetics (Bertamini et al. 2019). However, asymmetry is use of advanced annotation algorithms, fine mapping,
also observed in nature, as exemplified by the claws of fid­ and experimental validation. Future studies would benefit
dler crabs, the eyes of flatfish, and internal organs like the from larger sample sizes and ethnically diverse populations
heart, suggesting adaptive benefits in certain contexts to improve the generalizability of our findings.
(Duboc et al. 2015). Among these, brain asymmetry is par­
ticularly intriguing due to its connection to functional Materials and Methods
brain divisions and its proposed role in the evolution of in­
telligence (Wu et al. 2022). Various theoretical models Genotype Preprocessing in UK Biobank
have been proposed to explain specific facets of brain The UK Biobank database received ethical approval from
asymmetry (Ocklenburg and Gunturkun 2022). These in­ the National Research Ethics Service Committee North
clude the dextral/chance model for hand preference and West-Haydock (reference 11/NW/0382), and all proce­
language dominance (McManus 1985), the pathological dures adhered to the World Medical Association
left-handedness model (Satz et al. 1985), the right-ear ad­ Guidelines. All participants provided informed consent.
vantage model for left-hemisphere dominance in speech The cohort comprises individuals aged between 40 and
and language (Godfrey and Grimshaw 2016), and the 69 years, primarily of White British ancestry, with a
10
Evolution of Human Brain Left–Right Asymmetry · https://doi.org/10.1093/molbev/msad181 MBE
minority representing other ancestries. Approximately their connections is detailed in the supplementary note,
50,000 participants were analyzed using the Applied Supplementary Material online.
Biosystems UK BiLEVE Axiom Array by Affymetrix, while
the remaining around 450,000 participants were run on QTL Mapping
the Applied Biosystems UK Biobank Axiom Array, which
Our analysis first involved excluding subjects with discrep­
shares 95% marker content with the former (Bycroft
ancies between their self-reported (Field 31) and genetic­
et al. 2018).
ally inferred sex (Field 22001). We then applied linear
The genotype data we downloaded comprised
regression to remove the influence of covariates, including
93,095,623 imputed SNPs for 487,411 individuals. We con­
age (Field 31), age2, age × sex, age2 × sex, and the top 20
ducted the following preprocessing steps on the genotype
principal components (Field 22009-0.1 to 22009-0.20)
data: initially, we excluded SNPs with a minor allele fre­
that capture population genetic diversity. The residuals
quency (MAF) of less than 0.01 and an imputation INFO
obtained were used for subsequent QTL mapping.
score of less than 0.8. Subsequently, we retained SNPs
We computed the genetic relatedness matrix (GRM)
with a genotype call probability greater than 0.9 using
using the overlapping SNPs with LD-pruned HapMap3
the qctool. We also ensured that only biallelic SNPs were
SNPs for White and non-White British populations, in ac­
included. Finally, we excluded SNPs with a P value of less
cordance with a previous study (Jiang et al. 2019). The LD
than 10−6 based on Hardy–Weinberg equilibrium testing
pruning process was executed via PLINK (parameters:
and those with a genotype missing rate greater than 0.05
--indep-pairwise, window size = 1000 variant count, step
for both White and non-White British populations.
size = 100 variant count, pairwise r2 = 0.9 and MAF =
0.01). Thereafter, a sparse GRM was generated with
Definition and Preprocessing of Brain Asymmetry GCTA (–make-bK-sparse 0.05). We then undertook QTL
Trait mapping using GCTA (–fastGWA-mlm) with the sparse
GRM as an input, resulting in summary statistics for
Traits within the UK Biobank are designated by field num­
each asymmetry trait.
ber, instance number, and array number, formatted as
We determined significant SNPs at a threshold of
“field number-instance number.array number.” For in­
P < 5 × 10−8, removing SNPs with a MAF of less than
stance, the trait coded as “25565-2.0” pertains to field
0.01. Finally, PLINK was used to perform clumping analysis
25565, measured in instance 2, with an array number of
(r2 = 0.5 and kb radius = 250). Postclumping, we obtained
0. In this study, we collected data for all brain-related traits
the lead SNPs or index SNPs, each representing a QTL.
measured in paired brain regions (left and right hemi­
spheres). Each trait contains two instances: instance 2, re­
presenting the imaging visit (2014+), and instance 3, Replication Analysis
denoting the first repeat imaging visit (2019+). These in­ We defined three phases as follows: Phase 1 comprises
stances were treated as two separate time periods, aiding asymmetry traits measured during the first time period
in evaluating the impact of temporal variation on the iden­ for the White British group, identified by Field 22006,
tification of genetic variants. All traits analyzed in this and includes approximately 30,000 individuals with
study had an array number of 0. 8,895,704 SNPs. Phase 2 encompasses asymmetry traits
Using “L” to signify a trait measuring a specific brain re­ measured during the same period but for the non-White
gion in the left hemisphere and “R” to denote a trait British group. Phase 3 consists of asymmetry traits mea­
measuring the corresponding region in the right hemi­ sured during the second time period for the White
sphere, we defined brain asymmetry traits for each pair
√����������� British group, which includes around 1,000 individuals.
by arcsin((L − R)/ 2(L2 + R2 )) or
of these traits􏽰����������� Phases 2 and 3 are subsequently used as replication data
arcsin((Ls − Rs )/ 2(L2s + R2s )), as well as their absolute sets.
form, where “Ls” and “Rs” denote the standardized values Determining the non-White British group involved
of the traits “L” and “R,” respectively. As a result, we gener­ three steps. First, we identified individuals who reported
ated four asymmetry traits for each trait pair. We subse­ as non-White in Field 22006. Second, we estimated the
quently excluded asymmetry traits with a sample size of GRM for this group, obtained the sparse GRM, and re­
fewer than 1,000, as well as those traits where more than moved subjects exhibiting relatedness with more than
50% of the participants from the White British population 100 other individuals in the sparse GRM. These individuals
shared identical values. We carried out normalization were identified as having mixed genetic backgrounds.
transformations for each of these asymmetry traits, con­ Finally, subjects reporting mixed genetic backgrounds or
sidering White and non-White British populations separ­ those with missing values in Field 21000 (21000-0.0)
ately. The primary normalization was performed using were excluded. This second step was repeated to ensure
the R function “bestNormalize”; in cases where poor trans­ a relatively clear genetic background. As a result, the
formation was obtained, we substituted this function with non-White British group comprised approximately 3,000
“scale.” individuals.
A thorough mathematical review of the advantages of To evaluate the consistency between the discovery
our revised definitions over the traditional formula and (phase 1) and replication (phases 2 and 3) data sets for
11
Wang et al. · https://doi.org/10.1093/molbev/msad181 MBE
traits with QTLs in the discovery data set, we utilized three by the original authors through synteny comparison and
indexes. First, we calculated Pearson’s correlation (R) be­ protein family analysis, respectively. We contrasted the dis­
tween the β coefficients of the discovery data set and those tribution of the identified BA genes across these clades
of a replication data set for each trait, using the top 200 against the total gene distribution as a baseline. A binomial
SNPs obtained through clumping in the discovery data test was employed to gauge the significance of enrichment
set (–clump-p1 0.0001 --clump-p2 0.0001). Second, we as­ at each clade, setting a significance level at 0.05. This pro­
sessed the SCR between the β coefficients of the top 200 cedure was replicated, considering only genes significantly
SNPs in the discovery data set and a replication data set. up- or downregulated in brain tissues. For comparison be­
The significance was evaluated using a binomial test (ex­ tween ProteinHistorian’s results and those from GenTree,
pected success probability = 0.5). We combined all β coef­ we grouped all clades preceding Tetrapoda into an “an­
ficients of the top 200 SNPs for traits with QTLs to yield a cient” clade within ProteinHistorian. To address potential
single SCR. Third, we identified QTLs (lead SNPs) discov­ nonindependence issues between evaluations of different
ered in the replication data set that were initially identified clades, we conducted a stepwise binomial test. For in­
in the discovery data set. The significance of this identifica­ stance, in using ProteinHistorian, we first identified
tion was evaluated using a binomial test with an expected Bilateria as the most significant clade. We then removed
success probability equal to 0.05/n (Bonferroni correction), the genes at clade Bilateria from both the BA gene set
where “n” is the number of overlapping lead SNPs in a rep­ and the background gene set. Subsequently, we applied
lication data set. the same binomial test to the remaining clades to identify
the most significant among them. This process was reiter­
Gene Annotation ated until no additional significant clades were detected.
To annotate the associated genes for each of the lead or in­ Of the 216 BA genes we identified, 176 overlap with the
dex SNPs, we utilized both ANNOVAR and PhenoScanner. total background genes in GenTree, while 169 do so in
ANNOVAR accepts SNP locations as input, while ProteinHistorian.
PhenoScanner uses SNP identifiers. For the annotations
provided by ANNOVAR, we eliminated instances of
BSI of Gene Expression
multiple-gene annotation for a single SNP, including inter­
genic annotations. Using the Ensembl ID annotation informa­ Given a developmental stage and a specific species, the ex­
tion provided by the HGNC database (version 2023-3-2), we pression levels of a gene in six distinct organs (brain, heart,
translated the gene names given by ANNOVAR into Ensembl liver, kidney, ovary, and testis) of that species are repre­
IDs. As PhenoScanner provides a unique Ensembl annotation sented as a vector (x1 , x2 , x3 , x4 , x5 , x6 ). Each element of
for each SNP, we directly collected the Ensembl IDs it offered. this vector corresponds to one of these􏼠organs.􏼡The BSI
Ultimately, we combined the annotations from both soft­
of this gene is computed as BSI = log2 􏽐x61 , where
ware solutions due to the high degree of overlap in their x
i=2 i
results. xi represents the average gene expression of the focal
gene in the ith organ among the biological replicates of
Functional Enrichment Analysis this organ, x1 represents that in the first organ, that is,
To perform GO enrichment analysis, we employed the R 􏽐
the brain, and 6i=2 xi denotes the average expression level
package “clusterProfiler,” inputting Ensembl IDs and referen­ in the other organs.
cing “org.Hs,.eg.,db.” Subsequently, we utilized FUMA’s online To assess whether BA genes exhibit a significantly higher
platform to conduct tissue enrichment analyses, adhering to BSI in humans than in another species, we first kept the
default settings. The GO enrichment analysis underscored common organs without missing values and with expres­
significant gene enrichments in biological processes, cellular sion level satisfying the criterion RPKM > 1 in both species,
components, and molecular functions. FUMA’s tissue enrich­ and then we employed a one-tailed pairwise Wilcoxon test
ment analysis revealed considerable enrichments of BA genes to determine the P value. The robustness of the criterion is
in both upregulated and downregulated DEGs across various evaluated by comparing the results of it with those of an­
tissues. We identified significantly upregulated or downregu­ other two criteria, RPKM > 0.5 and RPKM > 1.5. The ana­
lated genes in brain tissues compared with nonbrain tissues lysis excludes the red junglefowl from the original study
using a two-tailed t-test, with a significance level of 0.05, ad­ (Cardoso-Moreira et al. 2019) due to the unavailability of
justed for multiple comparisons with Bonferroni correction. its Ensembl IDs from the Ensembl database. The cerebel­
Notably, of the 216 genes, 16 were not present in the results lum is not considered in the analysis as it was treated as
generated by FUMA. an independent organ in the original study, and only a
small fraction of our investigated traits relates to it.
Inferring Gene Origin by Macroevolution Analysis At stage 13, gene expression profiles are available for
We referenced two previous studies, GenTree and four organs in humans: the brain, heart, liver, and testis.
ProteinHistorian, which identified clade-specific origins These organs are also profiled in the other five species,
of genes. Each of these studies features key evolutionary making stage 13 more suitable for comparison than stage
clades, with genes that emerged at each clade mapped 14, where the number of organs varies between different
12
Evolution of Human Brain Left–Right Asymmetry · https://doi.org/10.1093/molbev/msad181 MBE
species pairs. Due to the consistent number of organs at When we applied this testing individually to each brain re­
stage 13, the BSI for most genes can be calculated consist­ gion for both donors and their average, we identified two
ently across all species, with only a few exceptions. regions for donor 1 (LiG-pest with P = 0.0099 and LOrG
Therefore, when creating the plots in figure 5, we used with P = 0.032), one region for donor 2 (ITG-mts with
the average BSI of genes obtained from comparisons be­ P = 0.014), and one region for their average (MTG-s with
tween humans and each of the other species. Notably, P = 0.028), all with a P value less than 0.05. However, these
we calculated the P values based on the BSI, which was de­ did not survive correction for multiple testing via the
termined using the common organs for each gene be­ Benjamini–Hochberg method. In this context, LiG-pest
tween species pairs. This approach ensures a rigorous represents the lingual gyrus, left/right, peristriate; LOrG
comparison. represents the lateral orbital gyrus; ITG-mts represents
the inferior temporal gyrus, left/right, bank of mts; and
MTG-s represents the middle temporal gyrus, left/right,
Comparing Asymmetry in Gene Expression: BA Genes superior bank of gyrus.
Versus Others
We obtained RNA-Seq data of two donors from the Allen
Brain Atlas (http://human.brain-map.org/static/download), Supplementary Material
which includes measurements for approximately 20,000 Supplementary data are available at Molecular Biology and
genes across around 30 brain regions (32 for donor 1 and Evolution online.
34 for donor 2) in both the left and right hemispheres.
These brain regions were identified using the annotation files
that were downloaded with the RNA-Seq data. We then cal­ Acknowledgments
culated the AL of gene expression between the left and right
We are grateful to X. Shen and J. Yang for technical assist­
hemispheres for each brain region, using the formula
√����������� ance. This work was supported by the National Natural
|L − R|/ 2(L2 + R2 ). This calculation is consistent with
Science Foundation of China (grant nos. 31970570 and
the plan-II of our asymmetry definitions, although it does
32070687) and the Shanghai Municipal Science and
not include an arcsine transformation. Theoretically, based
Technology Major Project (grant no. 2017SHZDZX01).
on this definition, the AL could range from 0 to 1.
When multiple measurements were made for a left or
right brain region, we used the average expression of these
replicates to provide a single gene expression value, omit­
Data Availability
ting any missing values. If a gene had missing values in ei­ Supplementary Material and the code used for analysis in
ther hemisphere for a specific pair of brain regions, the AL this study can be found on GitHub at https://github.com/
for that gene between those regions was also reported as a Jianguo-Wang/BAgenes.
missing value. To minimize bias from low expression va­
lues, we classified expression values in the original data
with a relative TPM less than 1e−7 as missing values. This References
accounted for 22% of the data for donor 1 and 21% for do­ Barbeito-Andres J, Bernal V, Gonzalez PN. 2016. Morphological
nor 2. Consequently, there remain 14 regions for donor 1 asymmetries of mouse brain assessed by geometric morphomet­
and 12 regions for donor 2 where nonmissing AL values ric analysis of MRI data. Magn Reson Imaging. 34:980–989.
Bein O, Reggev N, Maril A. 2020. Prior knowledge promotes hippo­
were obtained. campal separation but cortical assimilation in the left inferior
After conducting these calculations, we obtained the AL frontal gyrus. Nat Commun. 11:4590.
for BA genes and other genes across all pairs of brain regions. Bertamini M, Rampone G, Makin ADJ, Jessop A. 2019. Symmetry
We used a one-tailed Wilcoxon test to determine whether preference in shapes, faces, flowers and landscapes. PeerJ 7:e7078.
BA genes had a significantly higher AL than other genes, Briscoe SD, Ragsdale CW. 2019. Evolution of the chordate telenceph­
alon. Curr Biol. 29:R647–R662.
yielding the corresponding P value. A total of 174 BA genes Bycroft C, Freeman C, Petkova D, Band G, Elliott LT, Sharp K, Motyer
were subjected to this AL analysis. The gene identities were A, Vukcevic D, Delaneau O, O’Connell J, et al. 2018. The UK
determined by first transforming their Ensembl IDs into Biobank resource with deep phenotyping and genomic data.
Entrez IDs, using the annotation information provided by Nature 562:203–209.
the HGNC database (version 2023-3-2) and then aligning Byers JA, Waits L. 2006. Good genes sexual selection in nature. Proc
Natl Acad Sci U S A. 103:16343–16345.
them with the Entrez IDs measured in the Allen Brain
Capra JA, Williams AG, Pollard KS. 2012. ProteinHistorian: tools for
Atlas. Notably, the brain regions measured in the Allen the comparative analysis of eukaryote protein origin. PLoS
Brain Atlas represent only a small proportion of total brain Comput Biol. 8:e1002567.
regions, based on the provided annotation files. Cardinale RC, Shih P, Fishman I, Ford LM, Muller RA. 2013. Pervasive
We measured seven brain regions shared in both donors rightward asymmetry shifts of functional networks in autism
and found a significant correlation between the ALs for spectrum disorder. JAMA Psychiatry 70:975–982.
Cardoso-Moreira M, Halbert J, Valloton D, Velten B, Chen C, Shao Y,
each gene across these regions (Pearson’s R = 0.036, P < Liechti A, Ascencao K, Rummel C, Ovchinnikova S, et al. 2019.
2.2 × 10−16). We then averaged the AL for each gene across Gene expression across mammalian organ development.
these regions and conducted the same statistical testing. Nature 571:505–509.

13
Wang et al. · https://doi.org/10.1093/molbev/msad181 MBE
Corballis MC. 2002. From hand to mouth. The origins of language. Li X, Crow TJ, Hopkins WD, Gong Q, Roberts N. 2018. Human torque
Princeton (NJ): Princeton University Press. is not present in chimpanzee brain. Neuroimage 165:285–293.
Dean DC III, Planalp EM, Wooten W, Schmidt CK, Kecskemeti SR, Li P, Ensink E, Lang S, Marshall L, Schilthuis M, Lamp J, Vega I, Labrie
Frye C, Schmidt NL, Goldsmith HH, Alexander AL, Davidson RJ. V. 2020. Hemispheric asymmetry in the human brain and in
2018. Investigation of brain structure in the 1-month infant. Parkinson’s disease is linked to divergent epigenetic patterns in
Brain Struct Funct. 223:1953–1970. neurons. Genome Biol. 21:61.
Dehaene-Lambertz G, Spelke ES. 2015. The infancy of the human Lubben N, Ensink E, Coetzee GA, Labrie V. 2021. The enigma and im­
brain. Neuron 88:93–109. plications of brain hemispheric asymmetry in neurodegenerative
Duboc V, Dufourcq P, Blader P, Roussigne M. 2015. Asymmetry of the diseases. Brain Commun. 3:fcab211.
brain: development and implications. Annu Rev Genet. 49:647–672. Makowski C, van der Meer D, Dong W, Wang H, Wu Y, Zou J, Liu C,
Eckert MA, Vaden KI Jr, Iuricich F, Dyslexia Data C. 2022. Cortical Rosenthal SB, Hagler DJ Jr, Fan CC, et al. 2022. Discovery of gen­
asymmetries at different spatial hierarchies relate to phonologic­ omic loci of the human cerebral cortex using genetically in­
al processing ability. PLoS Biol. 20:e3001591. formed brain atlases. Science 375:522–528.
Finnerty JR, Pang K, Burton P, Paulson D, Martindale MQ. 2004. Manns M. 2021. It is not just in the genes. Symmetry (Basel). 13:
Origins of bilateral symmetry: Hox and Dpp expression in a sea 1815.
anemone. Science 304:1335–1337. Marie D, Roth M, Lacoste R, Nazarian B, Bertello A, Anton JL, Hopkins
Genikhovich G, Technau U. 2017. On the evolution of bilaterality. WD, Margiotoudi K, Love SA, Meguerditchian A. 2018. Left brain
Development 144:3392–3404. asymmetry of the planum temporale in a nonhominid primate:
Geschwind N, Galaburda AM. 1985. Cerebral lateralization: biologic­ redefining the origin of brain specialization for language. Cereb
al mechanisms, associations, and pathology: I. A hypothesis and a Cortex. 28:1808–1815.
program for research. Arch Neurol. 42:428–459. McManus IC. 1985. Handedness, language dominance and aphasia: a
Glasel H, Leroy F, Dubois J, Hertz-Pannier L, Mangin JF, genetic model. Psychol Med Monogr Suppl. 8:1–40.
Dehaene-Lambertz G. 2011. A robust cerebral asymmetry in Miao N, Lai X, Zeng Z, Cai W, Chen W, Sun T. 2020. Differential ex­
the infant brain: the rightward superior temporal sulcus. pression of microRNAs in the human fetal left and right cerebral
Neuroimage 58:716–723. cortex. Mol Biol Rep. 47:6573–6586.
Godfrey HK, Grimshaw GM. 2016. Emotional language is all right: Muntané G, Santpere G, Verendeev A, Seeley WW, Jacobs B, Hopkins
emotional prosody reduces hemispheric asymmetry for linguistic WD, Navarro A, Sherwood CC. 2017. Interhemispheric gene ex­
processing. Laterality 21:568–584. pression differences in the cerebral cortex of humans and ma­
Gonzalez PN, Vallejo-Azar M, Aristide L, Lopes R, Dos Reis SF, Perez caque monkeys. Brain Struct Funct. 222:3241–3254.
SI. 2022. Endocranial asymmetry in New World monkeys: a com­ Neubauer S, Gunz P, Scott NA, Hublin JJ, Mitteroecker P. 2020.
parative phylogenetic analysis of morphometric data. Brain Evolution of brain lateralization: a shared hominid pattern of en­
Struct Funct. 227:469–477. docranial asymmetry is much more variable in humans than in
Graic JM, Peruffo A, Corain L, Centelleghe C, Granato A, Zanellato E, great apes. Sci Adv. 6:eaax9935.
Cozzi B. 2020. Asymmetry in the cytoarchitecture of the area 44 Ocklenburg S, Gunturkun O. 2022. Cognitive and neurophysiological
homolog of the brain of the chimpanzee Pan troglodytes. Front models of brain asymmetry. Symmetry (Basel) 14:971.
Neuroanat. 14:55. Palomero-Gallagher N, Amunts K. 2022. A short review on emotion
GTEx Consortium. 2020. The GTEx Consortium atlas of genetic regu­ processing: a lateralized network of neuronal networks. Brain
latory effects across human tissues. Science 369:1318–1330. Struct Funct. 227:673–684.
Hain D, Gallego-Flores T, Klinkmann M, Macias A, Ciirdaeva E, Paracchini S, Diaz R, Stein J. 2016. Chapter two—advances in dyslexia
Arends A, Thum C, Tushev G, Kretschmer F, Tosches MA, genetics—new insights into the role of brain asymmetries. In:
et al. 2022. Molecular diversity and evolution of neuron types Friedmann T, Dunlap JC, Goodwin SF, editors. Advances in genet­
in the amniote brain. Science 377:eabp8202. ics. Academic Press. p. 53–97.
Hartwigsen G, Bengio Y, Bzdok D. 2021. How does hemispheric spe­ Park BY, Lariviere S, Rodriguez-Cruces R, Royer J, Tavakol S, Wang Y,
cialization contribute to human-defining cognition? Neuron 109: Caciagli L, Caligiuri ME, Gambardella A, Concha L, et al. 2022.
2075–2090. Topographic divergence of atypical cortical asymmetry and at­
Hibi M, Shimizu T. 2012. Development of the cerebellum and cere­ rophy patterns in temporal lobe epilepsy. Brain 145:
bellar neural circuits. Dev Neurobiol. 72:282–301. 1285–1298.
Hill H, Mirazón Lahr M, Beaudet A. 2023. Brain evolution and lan­ Parnell SE, Holloway HT, O’Leary-Moore SK, Dehart DB, Paniaqua B,
guage: a comparative 3D analysis of Wernicke’s area in extant Oguz I, Budin F, Styner MA, Johnson GA, Sulik KK. 2013.
and fossil hominids. Prog Brain Res. 275:117–142. Magnetic resonance microscopy-based analyses of the neuro­
Jiang L, Zheng Z, Qi T, Kemper KE, Wray NR, Visscher PM, Yang J. anatomical effects of gestational day 9 ethanol exposure in
2019. A resource-efficient tool for mixed model association ana­ mice. Neurotoxicol Teratol. 39:77–83.
lysis of large-scale data. Nat Genet. 51:1749–1755. Poeppel D, Assaneo MF. 2020. Speech rhythms and their neural
Kamat MA, Blackshaw JA, Young R, Surendran P, Burgess S, Danesh J, foundations. Nat Rev Neurosci. 21:322–334.
Butterworth AS, Staley JR. 2019. Phenoscanner V2: an expanded Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D,
tool for searching human genotype-phenotype associations. Maller J, Sklar P, de Bakker PI, Daly MJ, et al. 2007. PLINK: a tool
Bioinformatics 35:4851–4853. set for whole-genome association and population-based linkage
Kaplan G, Rogers LJ. 2021. Brain size associated with foot preferences analyses. Am J Hum Genet. 81:559–575.
in Australian parrots. Symmetry (Basel) 13:867. Saltoun K, Adolphs R, Paul LK, Sharma V, Diedrichsen J, Yeo BTT, Bzdok
Kong XZ, Mathias SR, Guadalupe T; ENIGMA Laterality Working D. 2023. Dissociable brain structural asymmetry patterns reveal un­
Group; Glahn DC, Franke B, Crivello F, Tzourio-Mazoyer N, ique phenome-wide profiles. Nat Hum Behav. 7:251–268.
Fisher SE, Thompson PM. 2018. Mapping cortical brain asym­ Satz P, Orsini DL, Saslow E, Henry R. 1985. The pathological
metry in 17,141 healthy individuals worldwide via the ENIGMA left-handedness syndrome. Brain Cogn. 4:27–46.
Consortium. Proc Natl Acad Sci U S A. 115:E5154–E5163. Schijven D, Postema MC, Fukunaga M, Matsumoto J, Miura K, de
Leroy F, Cai Q, Bogart SL, Dubois J, Coulon O, Monzalvo K, Fischer C, Zwarte SMC, van Haren NEM, Cahn W, Hulshoff Pol HE, Kahn
Glasel H, Van der Haegen L, Benezit A, et al. 2015. New human- RS, et al. 2023. Large-scale analysis of structural brain asymmet­
specific brain landmark: the depth asymmetry of superior tem­ ries in schizophrenia via the ENIGMA consortium. Proc Natl
poral sulcus. Proc Natl Acad Sci U S A. 112:1208–1213. Acad Sci U S A. 120:e2213880120.

14
Evolution of Human Brain Left–Right Asymmetry · https://doi.org/10.1093/molbev/msad181 MBE
Šestak MS, Domazet-Loso T. 2015. Phylostratigraphic profiles in zeb­ Wang K, Li M, Hakonarson H. 2010. ANNOVAR: functional annota­
rafish uncover chordate origins of the vertebrate brain. Mol Biol tion of genetic variants from high-throughput sequencing data.
Evol. 32:299–312. Nucleic Acids Res. 38:e164.
Sha Z, Pepe A, Schijven D, Carrión-Castillo A, Roe JM, Westerhausen Washington SD, Pritchett DL, Keliris GA, Kanwal JS. 2021.
R, Joliot M, Fisher SE, Crivello F, Francks C. 2021. Handedness and Hemispheric and sex differences in mustached bat primary audi­
its genetic influences are associated with structural asymmetries tory cortex revealed by neural responses to slow frequency mod­
of the cerebral cortex in 31,864 individuals. Proc Natl Acad Sci U S ulations. Symmetry (Basel) 13:1037.
A. 118:e2113095118. Watanabe K, Taskesen E, van Bochoven A, Posthuma D. 2017.
Sha Z, Schijven D, Carrion-Castillo A, Joliot M, Mazoyer B, Fisher Functional mapping and annotation of genetic associations
SE, Crivello F, Francks C. 2021. The genetic architecture of with FUMA. Nat Commun. 8:1826.
structural left-right asymmetry of the human brain. Nat Williams LZJ, Fitzgibbon SP, Bozek J, Winkler AM, Dimitrova R, Poppe
Hum Behav. 5:1226–1239. T, Schuh A, Makropoulos A, Cupitt J, O’Muircheartaigh J, et al.
Sha Z, Schijven D, Francks C. 2021. Patterns of brain asymmetry as­ 2023. Structural and functional asymmetry of the neonatal cere­
sociated with polygenic risks for autism and schizophrenia impli­ bral cortex. Nat Hum Behav. 7:942–955.
cate language and executive functions but not brain Won H, Huang J, Opland CK, Hartl CL, Geschwind DH. 2019. Human
masculinization. Mol Psychiatry. 26:7652–7660. evolved regulatory elements modulate genes involved in cortical
Shao Y, Chen C, Shen H, He BZ, Yu D, Jiang S, Zhao S, Gao Z, Zhu Z, expansion and neurodevelopmental disease susceptibility. Nat
Chen X, et al. 2019. Gentree, an integrated resource for analyzing Commun. 10:2396.
the evolution and function of primate-specific coding genes. Wu T, Hu E, Xu S, Chen M, Guo P, Dai Z, Feng T, Zhou L, Tang W, Zhan
Genome Res. 29:682–696. L, et al. 2021. Clusterprofiler 4.0: a universal enrichment tool for
Spring S, Lerch JP, Wetzel MK, Evans AC, Henkelman RM. 2010. interpreting omics data. Innovation (Camb). 2:100141.
Cerebral asymmetries in 12-week-old C57Bl/6J mice measured Wu X, Kong X, Vatansever D, Liu Z, Zhang K, Sahakian BJ, Robbins
by magnetic resonance imaging. Neuroimage 50:409–415. TW, Feng J, Thompson P, Zhang J. 2022. Dynamic changes in
Vallortigara G, Rogers LJ. 2020. A function for the bicameral mind. brain lateralization correlate with human cognitive performance.
Cortex 124:274–285. PLoS Biol. 20:e3001560.
Vanderhaeghen P, Polleux F. 2023. Developmental mechanisms Xiang L, Crow TJ, Hopkins WD, Roberts N. 2020. Comparison of sur­
underlying the evolution of human cortical circuits. Nat Rev face area and cortical thickness asymmetry in the human and
Neurosci. 24:213–232. chimpanzee brain. Cereb Cortex. 1–15.
Wan B, Bayrak S, Xu T, Schaare HL, Bethlehem RAI, Bernhardt BC, Zhao L, Matloff W, Shi Y, Cabeen RP, Toga AW. 2022. Mapping complex
Valk SL. 2022. Heritability and cross-species comparisons of hu­ brain torque components and their genetic architecture and phe­
man cortical functional organization asymmetry. Elife 11:e77215. nomic associations in 24,112 individuals. Biol Psychiatry. 91:753–768.

15