The draft genome of the grass carp (Ctenopharyngodon idellus) provides insights into its evolution

Articles

OPEN

The draft genome of the grass carp (Ctenopharyngodon idellus) provides insights into its evolution and vegetarian adaptation

©2015Nature America, Inc. All rights reserved.Yaping Wang1,11, Ying Lu2,11, Yong Zhang3,4,11, Zemin Ning5,11, Yan Li2, Qiang Zhao2, Hengyun Lu2, Rong Huang1, Xiaoqin Xia1, Qi Feng2, Xufang Liang6,7, Kunyan Liu2, Lei Zhang2, Tingting Lu2, Tao Huang2, Danlin Fan2, Qijun Weng2, Chuanrang Zhu2, Yiqi Lu2, Wenjun Li2, Ziruo Wen2, Congcong Zhou2, Qilin Tian2, Xiaojun Kang1,8, Mijuan Shi1, Wanting Zhang1, Songhun Jang1,9, Fukuan Du1, Shan He6,7, Lanjie Liao1, Yongming Li1, Bin Gui1, Huihui He1, Zhen Ning1, Cheng Yang1,8, Libo He1, Lifei Luo1, Rui Yang10, Qiong Luo10, Xiaochun Liu3,4, Shuisheng Li3,4, Wen Huang3,4, Ling Xiao3,4, Haoran Lin3,4, Bin Han2 & Zuoyan Zhu1The?grass?carp?is?an?important?farmed?fish,?accounting?for?~??6%?of?global?freshwater?aquaculture,?and?has?a?vegetarian?diet.??Here?we?report?a?0.9-Gb?draft?genome?of?a?gynogenetic?female?adult?and?a???.07-Gb?genome?of?a?wild?male?adult.?Genome?annotation?identified?27,263?protein-coding?gene?models?in?the?female?genome.?A?total?of?????4?scaffolds?consisting?of?573?Mb??are?anchored?on?24?linkage?groups.?Divergence?between?grass?carp?and?zebrafish?is?estimated?to?have?occurred?49–54?million?years?ago.?We?identify?a?chromosome?fusion?in?grass?carp?relative?to?zebrafish?and?report?frequent?crossovers?between?the??grass?carp?X?and?Y?chromosomes.?We?find?that?transcriptional?activation?of?the?mevalonate?pathway?and?steroid?biosynthesis?in?liver?is?associated?with?the?grass?carp’s?adaptation?from?a?carnivorous?to?an?herbivorous?diet.?We?believe?that?the?grass??carp?genome?could?serve?as?an?initial?platform?for?breeding?better-quality?fish?using?a?genomic?approach.Constituting a member of the Cyprinidae family and the only species of the genus Ctenopharyngodon, the grass carp Ctenopharyngodon idellus is one of the most important aquaculture species, having great commercial value and a worldwide distribution1 (Fig. 1a). Global pro-duction of cultured or farmed grass carp is approximately 4.6 million tons per year, accounting for 15.6% of global freshwater aquaculture production in 2011 (ref. 2). The completion of the zebrafish (Danio rerio) genome sequence3

has accelerated studies on the genomes of other members of the Cyprinidae family. In grass carp, progress included the construction of a genetic linkage map4 and the identifi-cation of 3,027 UniGene entries5 and 6,269 ESTs6. Such studies have enriched genome research on grass carp. Recent work has focused on genes involved in the immune system7,8, control of food intake9,10, and nutrition and growth11,12. However, the lack of a complete genome sequence has made it difficult to conduct an in-depth investigation of grass carp biology and breeding for better-quality fish. As a first step toward this goal, we report a draft genome sequence and tran-scriptomic analysis of grass carp, adding this important species to the other sequenced teleosts: cod13, fugu14, medaka15, tetraodon16, stickleback17 and zebrafish3. Taken together, this information pro-vides genomic insights into the evolutionary history of the grass carp and its unique adaptation to a vegetarian diet.RESULTSGenome?assemblyThe grass carp genome is composed of 24 pairs of chromosomes18,19 (2n = 48). Adopting a whole-genome shotgun sequencing strategy, we gener-ated approximately 132 Gb of Illumina sequence reads on genomic DNA isolated from the blood of a gynogenetic female adult grass carp and 136 Gb of reads from a wild, water-captured male adult (Supplementary Table 1 and Supplementary Note). We constructed the final assemblies of the female (0.90-Gb) and male (1.07-Gb) genomes using the modi-fied de novo Phusion-meta assembly pipeline, as previously described20 (Supplementary Fig. 1 and Supplementary Table 2). The draft genome of the female was fully annotated to mine genomic information (Fig. 1) and was applied to the anchoring of scaffolds on the genetic linkage

1State Key Laboratory of Freshwater Ecology and Biotechnology, Institute of Hydrobiology, Chinese Academy of Sciences, Wuhan, China. 2National Center for Gene Research, Shanghai Institute of Plant Physiology and Ecology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China. 3State Key Laboratory of Biocontrol, School of Life Sciences, Sun Yat-Sen University, Guangzhou, China. 4Guangdong Province Key Laboratory for Aquatic Economic Animals, School of Life Sciences, Sun Yat-Sen University, Guangzhou, China. 5Wellcome Trust Sanger Institute, Wellcome Trust Genome Campus, Hinxton, UK. 6Key Laboratory of Freshwater Animal Breeding of the Ministry of Agriculture, College of Fisheries, Huazhong Agricultural University, Wuhan, China. 7Freshwater Aquaculture Collaborative Innovation Center of Hubei Province, College of Fisheries, Huazhong Agricultural University, Wuhan, China. 8School of Computer Science, China University of Geosciences, Wuhan, China. 9College of Life Science, Kim Illinois Sung University, Pyongyang, North Korea. 10College of Plant Protection, Yunnan Agricultural University, Kunming, China. 11These authors contributed equally to this work. Correspondence should be addressed to Y.W. (wangyp@http://wendang.chazidian.com), H. Lin (lsslhr@http://wendang.chazidian.com), B.H. (bhan@http://wendang.chazidian.com) or Z.Z. (zyzhu@http://wendang.chazidian.com).

Received 4 April 2014; accepted 20 March 2015; published online 4 May 2015; doi:10.1038/ng.3280

Nature GeNetics

Articles

Figure 1 Assemblies and evolution of the grass carp genome. (a) Image of a grass carp adult. (b) Distribution of 55-mer frequency. The distribution of K-mer frequency in the reads was derived from libraries of short insert size (350–400 bp). Values for K-mers are plotted against the frequency (y axis) of their occurrence (x axis). The leftmost truncated peak at low occurrence (1–2) was mainly due to random base errors in the raw sequencing reads.

(c) Reconstructed phylogeny of 13 vertebrate

genomes. The dN/dS ratio of each branch is shown in blue. The numbers in black correspond to values of bootstrap support. The lamprey is used as an outgroup. Branch length is measured in expected substitutions per site. (d) Venn diagram of gene clusters for five selected vertebrate genomes. Each number represents the number of orthologous gene families shared by the indicated genomes.

a

b70

Frequency (million)

60504030201001

21

41

61

81

101

K-mer occurrence (K = 55)

MaleFemale

c

0.091

0.003

map, whereas the male genome was used to detect sequence variation between the male and female genomes.

The female genome assembly had scaf-folds with an N50 length greater than 6.4 Mb, 0.1and 90% of the assemblies were composed of

301 scaffolds, which were all greater than 179 kb in length (Table 1). Estimation of genome size by distribution of K-mer frequency showed that the female genome was about 891 Mb, close to the size of the assem-blies (Fig. 1b and Supplementary Note). We assessed the accuracy of the genome assembly by alignment of the scaffolds to 3,027 published UniGene entires5 and 11 BACs21 (Supplementary Figs. 2 and 3, and Supplementary Table 3), which indicated that the coverage by the ini-tial contigs and scaffolds was approximately 95% and 97%, respectively. Sequence errors were predominantly from insertions or deletions intro-duced by short-read assembly (Supplementary Table 3a).

We identified a total of 644,817 heterozygous SNPs and 66,101 short indels (10 nucleotides in length or less) in the female genome. In the male genome, we identified 1,465,819 SNPs and 166,867 short indels. The estimated overall heterozygous rates were approximately 0.9 and

table 1 Overview of assembly and annotation for the grass carp genome

Female

Total length

Length of unclosed gaps N50 length (initial contigs) N50 length (scaffolds) N90 length (scaffolds)

Quantity of scaffolds (>N90 length) Largest scaffold GC content

Quantity of predicted protein-coding genes Quantity of predicted noncoding RNA genes Content of transposable elements

Length of scaffolds anchored on linkage groups Quantity of scaffolds anchored on linkage groupsMale

Total length

N50 length (initial contigs) N50 length (scaffolds) N90 length (scaffolds)

Quantity of scaffolds (>N90 length) Largest scaffold

900,506,596 bp35,069,100 bp40,781 bp6,456,983 bp179,941 bp

301

19,571,558 bp

37.42%27,2631,57938.06%

573,471,712 bp (64%)

1141,076,149,922 bp

18,252 bp2,279,965 bp3,052 bp6,950

16,339,329 bp

©2015Nature America, Inc. All rights reserved.

0.133Grass carp1000.110Zebrafish

0.138Stickleback

0.0700.1711000.163Tetraodon

0.122100

0.171

1000.122Fugu

1000.0600.106Medaka100

0.087Cod0.111Anolis0.222

0.108Chicken0.128100

0.125100Human0.1290.160

1000.093Mouse

100

0.099Frog

Lamprey

d

Human

8,949

72

944

205

53

2552256

Grass carp7,871

Zebrafish7,6405333

60175175

165211

2

6765,7225515

13976389

Frog7,307

51

331533314

1011

129

Chicken7,148

2.5 polymorphisms per kilobase in the female and male genomes, respectively (Supplementary Table 4). Clearly, the wild male genome had a much higher heterozygosity rate, which caused a bimodality in the distribution of the K-mer frequency (Fig. 1b) and a shorter length for the assembled scaffolds.

Genome?annotation

We annotated a total of 27,263 protein-coding genes in the female genome. The evidence used in gene prediction included 27 Gb of RNA sequencing (RNA-seq) data from 6 tissues (embryo, liver, spleen, brain, kidney and head kidney), over 3,000 known UniGene entries and homologous gene information from zebrafish (Ensembl release 67; Supplementary Fig. 4, Supplementary Table 5 and Supplementary Data Set). We predicted 1,538 tRNA, 24 rRNA, 207 small nucleolar RNA (snoRNA), 136 small nuclear RNA (snRNA) and 444 microRNA genes in our annotation of noncoding RNA genes (Supplementary Tables 6 and 7) and 467,783 simple-sequence repeats (Supplementary Table 8). De novo repeat annotation indicated an overall repeat con-tent of 38%, in comparison to that of 43% in BACs (Supplementary Table 9). This proportion is less than the 52.2% repeat content observed in zebrafish3. This difference might be due to the exclusion of repetitive sequences located in unclosed gaps and on small frag-ments (<200 bp) of the grass carp assemblies. The majority of trans-posable elements found in the grass carp genome were type II DNA transposable elements, covering over 20% of the genome, similar to in the zebrafish genome3.

Using the published genetic linkage map of grass carp4, we anchored 114 scaffolds on the 24 linkage groups (Fig. 2a, Supplementary Table 10 and Supplementary Note), covering 573 Mb (64%) of the female assembly with 17,456 (64%) annotated genes localized. Gene synteny over the anchored scaffolds showed that most of the grass carp link-age groups had extensive collinearity with corresponding zebrafish chromosomes (Fig. 2b). Alignment of genes showed high synteny for grass carp and zebrafish, and up to 24,018 grass carp genes (88% of the 27,263 total genes) were located on syntenic blocks (Fig. 2c). Although this result was similar to a previous report4, we found two cross- chromosome arrangements for linkage groups 22 and 24. It is noteworthy

Nature GeNetics

Articles

that linkage group 24 aligned to zebrafish chromosomes 22 and 10 but not to any other grass carp linkage group. FISH analysis of grass carp chromosomes demonstrated that two grass carp markers aligning to

zebrafish chromosomes 10 and 22 were indeed located on the single linkage group 24 (Fig. 2d), explaining why the chromosome number is 25 in zebrafish but 24 in grass carp.

LG6

CID1503CID0515

CID0522CID0372CID0340(CID0119)CID0132CID1024CID1527CID0298(CID0214)CID0807CID0924CID0720

CID0833CID1504CID0306

CID0047

a

cM

LG1

CID0428

LG2LG3

CID0904

LG4

CID1031CID1517CID1538CID0796(CID1003)

LG5

CID1016

LG7

CID0899

LG8

CID0267CID0304CID0477CID0479

LG9

CID0044CID0291CID0556CID1516CID0754

LG10

CID0347CID0705

LG11LG12

CID0572

20

CID0696CID0029

CID1091CID0722(CID0114)CID0423

CID0972CID0818CID0050CID1539

CID1520CID0181CID1008CID0663

CID0142

CID0392CID0249CID0109CID0808CID1514CID0257

CID0103CID0059(CID0231)

CID0356CID1105CID0281CID0242CID1056CID0478

SNP0017CID0909

CID0395CID0139CID1085CID0121CID0255CID1526EST0004CID0658CID0212CID0161

40

CID0921CID1519CID0028CID1055CID0737CID0104CID1528CID0443

CID0578SNP0054CID0146(SNP0033)CID0452CID0349

CID0784CID0362

CID0190CID1067CID1506(CID1012)

CID0280CID1505CID1064CID0984CID0248

CID0357CID1058CID0441CID1532CID0634CID1525

CID0632CID0823CID0292SNP0034

CID0058CID0923CID0706

EST0006

60

CID0218CID0751(CID0225)

CID0465

©2015Nature America, Inc. All rights reserved.

CID0861

CID0855

80

5 Mb

CID0756

100cM

LG13

LG14

CID0593

LG15

CID0470CID1524

LG16

CID0123CID0276CID0442CID0025CID0487CID0177

LG17

CID0927CID1054CID0639CID0651CID0037CID0758CID0066CID0493(CID0388)(CID0483)CID0394SNP0002CID0283CID0321

LG18

CID0412CID0476(CID0825)CID0414SNP0040CID1535CID1111CID0735CID0259(CID0943)CID0763CID0745CID0055CID1009CID0270CID0323

LG19

CID0202CID0268CID0513CID0178CID320ACID320BSNP0053SNP0014

LG20

ClD1071CID0848

LG21

CID1529

LG22

CID0060CID0472CID0208CID0421CID0305CID1530

LG23

CID0176CID0615EST0002

LG24

CID0538CID0856CID0719(CID0689)(CID0671)CID1435CID1534CID1014CID0869

SNP0067SNP0015(CID0492)

CID0389CID1518CID0508CID1114CID0382

SNP0016

20

CID0436CID0051CID0367

CID0598(SNP0030)

CID0075CID532BCID1509(CID0619)CID0126SNP0035

CID1500CID0167

CID1531

CID0474

40

CID0870CID0314CID0195

CID0173CID0140CID0327

CID0416

60

b

Figure 2 Female scaffolds anchored on the genetic map. (a) The scaffolds were anchored Syntenicon a published consensus linkage map4. The zebrafish

Chromosomeblue lines indicate the length of each linkage

(n = 25) 73211913231811211416142015956242582171022

group (LG) to which the markers are mapped. Map distances between markers are depicted

cDR01DR02DR03DR04DR05DR06DR07DR08DR09dCID0538on a Kosambi cM scale. Orange bars represent

CID1435the anchored scaffolds. The black lines

linking markers and scaffolds show the locations of the markers on the scaffolds. The length of each scaffold is shown relative to a 5-Mb scale bar. (b) Syntenic relationship DR10DR11DR12DR13DR14DR15DR16DR17DR18between the zebrafish chromosomes and the grass carp linkage groups. Linkage group 22 is aligned to zebrafish chromosomes 2 and 15, and linkage group 24 is aligned to zebrafish

DR19DR20DR21DR22DR23DR24DR25chromosomes 10 and 22. (c) Gene collinearity

10 Mbbetween zebrafish and grass carp. The zebrafish

chromosomes are represented by blue blocks Zebrafish

chromosome(for example, DR01). The grass carp scaffolds

Grass carp(length > 50 kb) are represented by orange

scaffold

blocks. Aligned genes are connected by green

lines. The lengths of the chromosomes and scaffolds are shown relative to a 10-Mb scale bar. (d) FISH study of linkage group 24. The yellow marker CID1435 is located on the region aligned to zebrafish chromosome 10, and the red marker CID0538 is aligned to zebrafish chromosome 22. Scale bar, 5 µm.

Grass carp

LG(n = 24)

123456789101112131415161718192021222324

Nature GeNetics

A

rticles

Evolutionary?analysisTo examine grass carp evolution, we clustered the grass carp gene models with the genes from 12 other vertebrate genomes and used 202 single-copy genes with one-to-one correspondence in the different genomes to reconstruct a phylogenetic tree (Fig. 1c and Supplementary Note). As a species of the Cyprinidae family, grass carp had the closest 22relationship to zebrafish. According to the TimeTree database, the estimated divergence time between zebrafish and grass carp was around 49–54 million years ago (Supplementary Table 11). Most of the selected teleostei genomes showed similar selection pressures, according to calculated dN/dS values (the ratio of the rate of non-synonymous substitution to the rate of synonymous substitution).We determined gene families using the TreeFam database23 (Supplementary Note). We performed a five-way comparison among the gene families of a representative mammal (human), bird (chicken), amphibian (frog) and two fishes (zebrafish and grass carp) to quantify the shared or species-specific families present in each genome (Fig. 1d). Zebrafish and grass carp shared 7,227 families, ©2015Nature America, Inc. All rights reserved.Carbohydrate metabolism Carbohydrate metabolism, citrate cycle (TCA cycle)Energy metabolism

Lipid metabolismLipid metabolism, glycerophospholipid metabolismNucleotide metabolismAmino acid metabolismMetabolism of other amino acidsGlycan biosynthesis and metabolismGlycan biosynthesis and metabolism, glycosaminoglycan biosynthesisMetabolism of cofactors and vitaminsMetabolism of terpenoids and polyketidesBiosynthesis of other secondary metabolitesXenobiotics biodegradation and metabolismTranscriptionTranslationFolding, sorting and degradationReplication and repairMembrane transportSignal transductionSignal transduction, PI3K-Akt signaling pathwaySignal transduction, Hippo signaling pathwaySignaling molecules and interaction

Signaling molecules and interaction, ECM-receptor interaction

Transport and catabolism

Cell motility

Cell growth and death

Cellular community

Cellular community, focal adhesion

Cellular community, adherens adhesion

Immune system

Endocrine system

Circulatory system

Digestive system

Excretory system

Nervous system

Sensory system

Development

Environmental adaptation

Cancers: overview

Cancers: specific types

Immune diseases

Neurodegenerative diseases

Substance dependence

Cardiovascular diseases

Endocrine and metabolic diseases

Infectious diseases: bacterial

Infectious diseases: viral

Infectious diseases: parasiticmore than the 5,772 families shared by all 5 vertebrate species. Of the 10,184 families identified, 7,171 (70%) carried the same number of gene members in grass carp and zebrafish. Specific com-parison of the Hox24,25, Sox26 and Toll-like receptor27 gene clusters among human, medaka, zebrafish and grass carp indicated that the zebrafish and grass carp genomes carried an identical copy number for most subfamilies (Supplementary Figs. 5–7 and Supplementary Table 12). We determined the number of human, zebrafish and grass carp gene members in each family (Supplementary Table 13). The 1,047 families in the class having many grass carp members relative to one human member were composed of 2,658 grass carp genes and 1,047 human genes, with an average ratio of 2.53 grass carp genes to one human gene. Interestingly, the 832 families in the class having many zebrafish members relative to one human member consisted of 2,077 zebrafish genes and 832 human genes, with nearly the same average ratio of 2.50 zebrafish genes to one human gene. It was suggested that the grass carp genome underwent a whole-genome duplication similar to zebrafish after the teleost radiation28.We estimated the expansion and contrac-tion of gene families to examine their evolu-tionary history in comparison to the zebrafish, stickleback, tetraodon, fugu, medaka and cod genomes (Supplementary Fig. 8). The signifi-cantly expanded families in grass carp included many immune-associated functional domains (P < 0.001; Supplementary Table 14), consist-ent with the adaption of grass carp to variable environments. Among the 10,184 gene fami-lies generated, 2,346 included teleost-specific duplications in zebrafish or grass carp as determined by comparison of the number of gene copies within each family. Of the gene families involved in the teleost-specific duplications, 695 and 295 showed evidence of undergoing a grass carp–specific duplication (GCSD) or a zebrafish-specific duplication (ZSD), respectively (Supplementary Fig. 9 and Supplementary Note), with additional gene duplications found in grass carp and zebrafish. The 695 grass carp families con-tained 2,561 genes, whereas the 295 zebrafish families consisted of 1,029 genes. We annotated all of these genes using the KEGG29 (Kyoto Encyclopedia of Genes and Genomes) path-way database. Functional analyses of these pathways indicated that genes involved in the ZSD were mainly composed of immune-related genes. Comparably, the grass carp genes involved in the GCSD were not only associated with immune-related genes but also with development-related genes (Fig. 3 and Supplementary Table 15) and were Figure 3 Distribution of GCSD- and ZSD-related genes by pathway. All of the genes were involved in the GCSD or ZSD. Pathways were determined by searching the KEGG pathway database. The x axis indicates the numbers of genes involved in each pathway. Pathways with a label in blue highlight metabolic processes that

are potentially important to development or

diet adaptation.Number of genes involved in the ZSD or GCSD

Nature GeNetics

Articles

involved in cell proliferation and differentiation (for example, the focal adhesion pathway and the extracellular matrix (ECM)-receptor interac-tion pathway30,31), nutritional homeostasis (for example, the protein diges-tion and absorption pathway32,33) and organ size control (for example,

34,35the Hippo signaling pathway). Comparison of genes involved in the

overview maps of metabolism (reference map ko01100 of the KEGG database) also indicated that grass carp genes involved in the GCSD clustered in carbohydrate metabolism and nucleotide metabolism

(Supplementary Fig. 10a), whereas the zebrafish genes only clustered

in nucleotide metabolism (Supplementary Fig. 10b). These results indicate that the GCSD was important for adaptation to a vegetarian diet and for some developmental characteristics of grass carp.

a

b

A?potential?sex-determination?mechanism

By comparison of the assemblies for the male and female grass carp, we identified 206 contigs with a total length of 2.38 Mb that were

carried by the male adult but not by the gyno-genetic female (Supplementary Fig. 11 and 46 d

Fed with116 d–FHTSupplementary Tables 16 and 17). We con-chironomid larvaeFed with

firmed each contig by PCR-based sequenc-duckweed

ing. We also performed PCR amplification of these regions in an extended group of 24 male and 24 female individuals, identify-ing frequent chromosome crossovers between 116 d

Still fed with

the X and Y chromosomes in grass carp chironomid

(Supplementary Fig. 12). Sex in grass carp larvae

may be determined not by an entire chromo-some but by a few critical genes. Noticeably, Days after hatchingwe identified a male genome–specific probe

that mapped to one of the contigs (probe

GlycolysisMevalonate pathway

184 in Supplementary Fig. 12). We did not in terpenoid backbone biosynthesis

3-hydroxy-3-methyl-glutaryl-find a sequence alignment of this region to

CoA1.1.1.88Mevalonate

any other vertebrate genome, suggesting its Acetyl-CoA

2.3.3.10

unique origin in grass carp.

Gene modeling showed that the male-specific contigs mainly contained genes with domains related to the immunoglobulin

V–set, ABC transporter, proteasome subunit and NACHT domains (Supplementary Mevalonate-Table 18). Alignment of these predicted genes 5Pto the female gene model set showed that 40 genes had homologs in the female genome,

of which 22 clustered on linkage group 24 Lanosterol(Supplementary Fig. 13). Gene collinear-ity and FISH analysis demonstrated that zebrafish chromosomes 10 and 22 fused 4,4-dimethyl-cholesta-to form a single chromosome—linkage 8,14,24-trienol

14-demethyl-lanosterol

©2015Nature America, Inc. All rights reserved.

degradation

Primary bile acidbiosynthesis

Figure 4 Characterization of gene expression during the FHT period. (a) Design of the FHT experiments. At 46 d after hatching, fish not undergoing FHT (fed with chironomid larvae) were collected as sample “46 d”. During the period from 46 to 116 d after hatching, the fish were divided into two groups—one fed with duckweed and the other still fed with chironomid larvae—which were collected at 116 d after hatching as sample “116 d– FHT” and sample “116 d,” respectively.

(b) Activation of the mevalonate pathway and steroid biosynthesis. The blue and orange arrows indicate the reaction steps in different pathways. The number in each rectangle shows the EC code of the enzyme catalyzing that transfer. Red codes indicate enzyme genes with significantly increased expression after the FHT (q < 0.001). Gene expression as measured by quantification of transcription levels (reads per kilobase of exon model per million mapped reads, RPKM45) is shown in the histograms. Compound names are shown beside the corresponding circles.

Nature GeNetics