Genome-Wide Association Analysis of Late Blight Resistance Traits in Potato Germplasm Resources

Background: Potatoes are dicotyledonous plants of the genus Solanum, family Solanaceae, and contain large amounts of starch, proteins, and trace elements required by the human. Potato late blight is the main disease hindering potato production. In this study, Phytophthora infestans were used to quantify late blight resistance in 284 germplasm resources, and resistance genes were mined through genome-wide association analysis. Results: The results showed that among the 284 potato germplasm resources, 37 showed immunity, 15 were highly resistant to late blight, 30 were moderately resistant to late blight, 107 were moderately susceptible to late blight, and 95 were highly susceptible to late blight. Through screening and ltering, 22,489 high-quality single-nucleotide polymorphisms (SNPs) and indels were obtained. Through population structure analysis and principal-component analysis, 284 germplasm resources were divided into eight subgroups, which was consistent with the results of the phylogenetic tree analysis. The genetic diversity index of the 284 potato germplasm resources was 0.2161, and the differentiation index of each subgroup was 0.0251-0.1489. A mixed linear model was built to perform an association analysis on the diameter of the lesions identied from isolated leaves of potato affected by late blight. The genes within 100 kb of both sides of the obtained signicant SNP loci were searched and functionally annotated, and 18 candidate genes were obtained. Twenty-two candidate genes were obtained from the association analysis of disease resistance grade. Conclusions: 284 potato germplasm resources were used to identify for Phytophthora infestans resistance. The potato germplasm resources were divided into 8 subgroups by population structure analysis, and the main differentiation among subgroups was moderate. Candidate genes were mined by genome-wide association analysis.The results of this study provides the foundation for the genetic improvement of potato varieties resistant to late blight.


Background
Potato (Solanum tuberosum L.) is one of the most important food crops in the world, its total output ranking third in the world [1]. Potato late blight is a fungal disease caused by Phytophthora infestans, and it is also the most serious disease in potato production [2]. The spores and sporangia of Phytophthora infestans can be spread widely and quickly through water and air currents and infect plant tissues within 7 to 10 days [3,4]. It causes dead leaves and rotten tubers, which seriously affect the yield and quality of potatoes [5]. Outbreaks of late blight have caused more and more genotypes to become susceptible to late blight, and it often leads to the decline or loss of resistance to late blight in resistant varieties in production. Therefore, pathogens causing late blight are considered to have the highest evolutionary potential, the greatest risk, and the most ease at overcoming a single or a few disease resistance genes [6].
In traditional agriculture, the use of protective agents or systemic fungicides is the main method of controlling late blight [7]. However, pesticides not only pollute the environment but also accelerate the development of resistance and the mutation of pathogens. Transgenic technology can quickly obtain disease-resistant varieties, but the adoption rate of genetically modi ed crops is still very low. In addition, the application of transgenic technology to food security has always been controversial [8].
The concept of genome-wide association study (GWAS) was proposed by Risch (1996) [9]. Hansen (2001) rst reported the application of a GWAS to the growth properties of wild beet [10]. Followed by Aranzana (2005) veri ed the feasibility of GWAS in Arabidopsis thaliana by studying its owering time and pathogen resistance [11]. This was the rst complete genome-wide association analysis in plants. With the advancement of statistical models and sequencing technologies, GWAS has gradually become an important tool for analysing the genetic basis of complex traits in plants. GWAS has been done in rice [12,13], corn [14], rape [15], sorghum [16], barley [17], and Arabidopsis [18], leading to the identi cation of multiple genes related to important agronomic traits. GWAS uses linkage disequilibrium (LD), that is, two loci in the genome with more or less statistical association in their sequences due to recombination history, to link genotype with phenotype. GWAS overcomes several shortcomings of quantitative trait locus mapping in that GWAS provides higher resolution, is less time-consuming, requires fewer resources, and gives more consideration to allelic diversity [19,20,21].
LD means that in a speci c population, the probability of co-occurrence of a speci c allele located at a certain locus and an allele of another locus is greater than that of two alleles randomly distributed in the population. It determines the accuracy of the association analysis and the number and density of molecular markers used. Researchers can use the LD decay distance to calculate the number of markers required for GWAS, then calculate the detection e ciency and accuracy of the GWAS: GWAS marker amount = genome size/LD decay distance. The selection of the population (the selection of germplasm materials), the hierarchical structure of the population, and the genetic drift of the population affect the accuracy of the association analysis.
With the advancement of high-throughput sequencing technology, GWAS has become an important means to identify the relationship between the target phenotypic traits and genetic markers or genes, as well as to detect allele loci that have a strong role in controlling the related traits [22]. In this study, a GWAS was performed on the late blight resistance results of 284 isolated samples of potato leaves to evaluate the genetic diversity, genetic structure, and genetic relationships within the experimental population and to screen the candidate genes related to late blight resistance of potato, thus laying the foundation for genetic methods to improve late blight resistance in potato.

Genotype analysis of potato population
All SNPs and indels were ltered for minor allele frequency (MAF) > 0.05 and Hardy-Weinberg equilibrium P value > 0.001, and 20,382 high-quality SNPs and 2,107 indels were obtained. Annotation of the high-quality SNPs and indels showed that 18,683 (83.07%) were located in intergenic regions; 3,806 (16.92%) were located in the gene regions of the genome, of which 951 were located in un-transcribed regions, 2,796 were located in introns, and only 1,682 SNPs were in coding regions. In the coding regions, 771 SNPs produced silent mutations and 911 SNPs produced missense mutations, for ratio of 1.18 (Table 1). Identi cation and analysis of the late blight resistance of potato leaves in vitro Among the 284 materials tested, 37 germplasm resources were asymptomatic or had only necrotic spots at the inoculation site, so they had a disease severity grade of 0; 15 germplasm resources had a lesion diameter (d) ≤ 5 mm, no chlorotic halo at the edge of the lesion, and thus a disease severity grade of 1; 30 germplasm resources had 5 mm < d ≤ 10 mm, the lesion spot was water-soaked with chlorotic halo on the edge, and so the disease severity grade was 2; 107 germplasm resources had 10 mm < d ≤ 20 mm, a water-soaked lesion with a chlorotic halo on the edge, and so a disease severity grade of 3; 95 germplasm resources had d > 20 mm, a lesion spot covered with a uniformly thick layer of mildew, and thus a disease severity grade of 4 (Table S1).

Analysis of population structure
The ADMIXTURE software was used to analyse 22,489 high-quality SNPs and Indels, the largest cluster subgroup value (K) was assumed to be each integer from 1-12, and the cross-validation (CV) error of each K value was calculated (Fig. 1A). When K was 1-4, the CV error gradually increased. When K was greater than 4, the CV error dropped rapidly to a nadir at K = 8, and for K > 8 it gradually increased. Therefore, K = 8 was optimal, that is, the entire potato population was divided into eight subgroups.
PCA was performed using all high-quality SNPs and indels. The calculation and analysis process were performed by R software (Fig. 1C). After the analysis was completed, plots were generated by R. For plotting, the eight subgroups inferred by ADMIXTURE software were used for grouping. The results showed that the eight subgroups could basically be distinguished on the PC1 axis, and the clustering results were consistent with the population structure division. Potatoes are native to the Andes of South America, and the history of arti cial cultivation can be traced back to southern Peru from 8,000 to 5,000 BC.
According to the Q value of each material in these eight subgroups, each material was classi ed into the subgroup with the largest Q value (Fig. 1B). Subgroup 1 to Subgroup 8 had 11, 33, 25, 16, 12, 30, 51, and 86 germplasm resources, respectively. The distribution of the eight subgroups showed difference on the PC1 axis, and the clustering results were consistent with the population structure division. The eight subgroups of potato could not all be clustered together on the phylogenetic tree.
(A) The population structure of 284 potato materials was analysed by using ADMIXTURE software. CV error was calculated when K = 1-12. (B) When K = 8. In this population structure, each individual is represented by a line with eight different colours. According to the proportion of colours, which subgroup the variety belongs to can be inferred. (C) PCA was performed on all 284 potato samples with high-quality polymorphic loci. Each dot represents a sample.
The neighbour joining method was used to construct a phylogenetic tree, and the tree diagram was drawn with iTOL software to explore the genetic relationships between the 284 potato germplasms. Overall, the clustering results were consistent with the division of the population structure: subgroup 1, subgroup 2, and subgroup 6 clustered together well, while samples of other subgroups could be clustered together, and there was certain cross-over between samples (Fig. 2).

Genetic diversity revealed by SNP markers
According to the 22,489 high-quality SNPs and indels data, the genetic diversity (π) of all 284 potato germplasm resource was 0.2161, and the genetic diversity index of eight subgroups was between 0.1638 and 0.2502. Among them, subgroup 8 had the lowest genetic diversity index (0.1638), and subgroup 6 had the highest (0.2502) ( Table 2). These data show that there was rich genetic diversity in the 284 potato germplasm resources.

GWAS analysis of late blight resistance of potato germplasms
With the obtained SNPs and indels, the mixed linear model analysis was used to perform correlation analysis on the lesion spot diameter identi ed on the isolated leaves affected by late blight (Fig. 3) and disease resistance grade (Fig. 4). For the lesion diameters, P < 4.4 × 10 − 7 was set as the threshold to screen signi cant loci, and 18 candidate genes were obtained after annotating genes located at or near the signi cant loci. For the disease resistance grade, P < 5 × 10 − 2 was set as the threshold to determine the signi cant loci, and 22 candidate genes were obtained after annotating genes located at or near the signi cant loci.

Potato genome-wide LD
The LD-decay curve was obtained by analysing the LD of all 284 potato germplasm resources using the 22,489 SNP markers obtained from the whole genome (Fig. 5). The results show that LD decreased as the physical distance between SNPs increased. When taking the coe cient of determination r 2 = 0.1 as the decay threshold, within this population the decay distances of subgroup 3 and subgroup 8 were approximately 0.9 kb, those of subgroup 5 and subgroup 7 were approximately 1 kb, those of subgroup 2 and subgroup 4 were approximately 1.1 kb, that of subgroup 1 and subgroup 6 were approximately 1.2 kb, and that of the entire population was approximately 0.9 kb. All of these are far lower than the LD-decay distances of cultivated rice (123 kb), cultivated soybean (133 kb), cultivated corn (30 kb), and cultivated cassava (8 kb) and slightly smaller than that of the maize inbred line population (1.5 kb) [23].

Candidate genes
Genome-wide association analysis of disease severity (lesion spot diameter) We annotated the genes located in the signi cant loci and within 100 kb around these loci, and 18 candidate genes were found. The detailed information is shown in Table 4. Among these genes, PGSC0003DMG400028682 encodes chitinase 1, which is related to immune response; PGSC0003DMG400036902 encodes a glycine-rich protein involved in the defence ability of the cell wall; and PGSC0003DMG400027651, PGSC0003DMG401013892, and PGSC0003DMG400015828 encode ethylene receptor 2, ethylene response transcription factor 7, and ERF transcription factor 5, which are all related to the response to ethylene. Ethylene can coordinate with jasmonic acid during invasion of pathogens. Protein kinases such as those encoded by PGSC0003DMG400000043, PGSC0003DMG400016913, PGSC0003DMG400044850, PGSC0003DMG400007634, and PGSC0003DMG400006739 can catalyse protein phosphorylation and are regulatory factors mediating signal transduction in response to external stimuli. PGSC0003DMG400031455 and PGSC0003DMG400047159 encode late blight resistance proteins, and PGSC0003DMG400031878 encode NBS-LRR resistance proteins, which are directly related to potato late blight resistance. Among them, NBS-LRR proteins are the most important late blight resistance proteins. PGSC0003DMG400023057 and PGSC0003DMG400016899 are leucine-rich repeat sequence family proteins that may be involved in the speci c recognition of pathogenic effectors.  Expression patterns of candidate genes Four candidate genes were randomly selected to verify their gene expression patterns in late blight-resistant varieties A1, CIP10-1, and 0422 − 19 and susceptible varieties D8, UK7, and FAVORITA by qRT-PCR.The green leaves around the plaque, after inoculation with Phytophthora infestans in vitro for ve days, and leaves without inoculated were used. Then their relative expression levels were calculated (Fig. 6). The results showed that most of the candidate genes were up-regulated after inoculation. The expression levels of PGSC0003DMG400028682 and PGSC0003DMG401013892 in resistant varieties were higher than those in susceptible varieties, while the expression levels of PGSC0003DMG400000043 in susceptible variety UK7 were higher than those in resistant variety 0422 − 19, and PGSC0003DMG400006739 in susceptible variety Favorita was higher than that in resistant variety Favorita Sexual variety A1. The results showed that the resistance to late blight was a quantitative trait controlled by multiple genes.

Discussion
In our study, we evaluated the resistance to Phytophthora infestans in 284 Potato Germplasms. Of these, 37 Potato Germplasms showed the immunity. Population structure analysis was used to assign the 284 potato germplasms into 8 subpopulations. The high-quality SNP markers revealed the moderate level in the main differentiation among the eight subpopulations. At the same time, genome-wide association analysis was carried out to identify the signi cant loci for late blight resistance traits in germplasm resources and screen candidate genes.

Identi cation and analysis of late blight resistance of Germplasm Resources
In recent years, the plethora of reports presented the identi cation of Potato Resistance to late blight; however, there are some differences in the identi cation results. The mixed strains selected in this study have yet to be determined at physiologic race level. With our studies for many years in the laboratory (Ref or unpublished data), we discovered that the resistance of potato leaves to Phytophthora infestans in vitro was basically consistent with that at the adult stage, and therefore, the identi cation results of detached leaves could represent the resistance at the adult stage. Feiwuruita is a high susceptible late blight variety, which is again consistent with our result [24]. It is speculated that identi cation of late blight resistant germplasm resources and discovery of durable resistant materials will be an important issue in breeding the high-quality late blight resistant potato varieties in near future. In this study, 37 materials showed immunity and 15 germplasms conferred highly resistance to late blight disease which provided the valuable materials for late blight resistant potato breeding.

Analysis of potato population structure and genetic diversity
In this study, 284 potato germplasm resources were used for the association analysis of resistance to late blight, 137 of which were from the International Potato Center (CIP). By testing genetic markers for Hardy-Weinberg equilibrium and suballelic frequency ltering, 22,489 high-quality polymorphic loci were obtained. The strati cation of population structure and the uneven distribution of alleles are important reasons for false associations between genotypes and traits [25]. The population structure analysis of 284 potato germplasm resources in this study found that when K = 8, the CV error value was the smallest. Therefore, the 284 potato germplasm resources were divided into eight subgroups. The genetic diversity index of the potato population was 0.2161, indicating that there was abundant genetic diversity. The F ST among subgroups was mostly between 0.05 and 0.15, and the differentiation between subgroups was mostly moderate, indicating that there was a certain degree of differentiation in germplasm resources, but the degree of differentiation was not high.

Analysis of candidate genes
Late blight is a serious disease worldwide, threatening the potato industry and food security. Late blight resistance potato breeding is very important in China and abroad, and breeding to have multiple disease resistance genes in the same variety is an important strategy to prevent late blight. So far, 11 broad-spectrum resistance genes (R1-R11) have been discovered, and these 11 major R genes have been successfully located on the potato genetic map [26,27]. Combinations of different disease resistance genes can provide ideal late blight resistance [28]. Therefore, optimizing the known disease resistance gene combinations, making full use of R disease resistance genes with broad-spectrum resistance characteristics, and discovering new broad-spectrum and longer-lasting disease resistance genes from the abundant potato resources are the effective means for the cultivation of resistant potato varieties in the future [29]. The abundant wild resources of potato are the source of R genes. Potato contains many genes encoding cytosolic nucleotide binding site-leucine-rich repeat (NBS-LRR) resistance proteins [30]. The R gene can be introduced into cultivated varieties by crossing conventional varieties with wild species containing this gene, thus helping cultivated varieties to achieve durable resistance to late blight [31]. In the abundant potato resources, there are many unknown late blight resistance genes waiting to be discovered.
In this study, 18 and 22 candidate genes, respectively, were found by the whole-genome association analysis of the diameter of the lesion spots and the disease severity grade identi ed from isolated potato leaves affected by late blight in vitro. Among them, the candidate gene encoding chitinase 1 gene may be involved in the immune response, and chitinase can improve the resistance of plants to fungi [32,33]. The whole-genome association analysis of lesion diameter and disease resistance grade found that three and ve candidate genes were associated with ethylene response, respectively. Ethylene can coordinate with jasmonic acid to regulate and function together during pathogen invasion [34]. The correlation analysis of lesion diameter found that two late blight resistance proteins and one NBS-LRR resistance protein may be directly related to potato late blight resistance. NBS-LRR proteins are the most important late blight resistance proteins and are characterized by the same conserved structure containing an N-terminal leucine zipper or coiled coil, a nucleotide-binding site, and leucine-rich repeats. The cloned anti-late blight genes contain several highly conserved structures, such as a phosphate-binding domain (Ploop), a kinase-2 group, or a GLPL group, in the NBS region [35,36]. This study provides genetic resources for follow-up research and lays the foundation for the genetic improvement of potato resistance to late blight.

Conclusions
The result showed that 37 potato germplasm resources were immuned to Phytophthora infestans and 15 were highly resistant to Phytophthora infestans. Eight subgroups were categorized according to population structure analysis. There was abundant genetic diversity among the subpopulations mainly with moderate differentiation.
The whole genome association analysis revealed that there were 18 candidate genes in the lesion diameter and 22 candidate genes in disease resistance grade.

Page 12/19
Sample collection The 284 potato germplasm resources used for this GWAS were collected from the Qinghai Plateau Potato Experimental Station, including 103 from China, one from Australia, three from Belarus, ve from Canada, ve from the United Kingdom, 133 from the international Potato Center (CIP), two from Israel, one from the Netherlands, ve from Russia, two from the United States, and 24 from unknown sources. The library label of these samples were recorded in Table S1. All materials were stored in Academy of Agriculture and Forestry Sciences, Qinghai University (E101°, N36°). No permission was required in collecting the plants. In this study, Fang Wang was responsible for the planting and identi cation of these samples.
Quanti cation of resistance to late blight in potato germplasm resources Preparation of microbe mixture The "Northwest Mixed Microbes" containing mixture of pathogenic microbes causing late blight was used for inoculation. The northwest mixture was composed of ve strains of "1352", "1353", "1835", "1837" and "1839", all of which were presented by Professor Jiehua Zhu of Hebei Agricultural University. The surface of the susceptible potato pieces was sterilized with a 95% alcohol ame, cut into slices approximately 0.5 cm thick, and placed in a petri dish covered with moist lter paper. After the potato slices were inoculated with the pathogen, they were placed in an 18 °C incubator and incubated in the dark for the recovery and ampli cation of the microbes. After the sporangia were produced in large quantities (approximately 3 to 5 days), the sporangia were collected, put into sterile water, ltered through a nylon mesh to remove the hyphae, and made into a sporangia suspension, which was placed in a refrigerator at 4 °C for 2 to 3 h. After the zoospores were released, the sporangia and empty sporangia shells were removed by ltering through a steel mesh with a pore size of 12 µm, and then the concentration was adjusted to 5 × 10 4 /mL with a haemocytometer for future use.

Indoor inoculation and identi cation
The varieties to be tested were sown in the experimental eld shed of Qinghai Plateau Potato Experimental Station (E101°, N36°). Each variety was sown in one row. The growth temperature was between 18 °C and 25 °C.
Natural light and timely fertilization, watering, and weeding were provided. When the seedlings grew to have 7-10 compound leaves, the leaves were picked for in vitro inoculation and identi cation indoors.
Two layers of sterilized lter paper were spread in a humidi ed transparent box, and a certain amount of sterile water was added for moisturizing. The leaves of the potato variety to be tested were placed on the lter paper with the back of the leaves facing upwards, and a micropipette was used to draw up 15 µL of the zoospore suspension and add the liquid dropwise onto the back of each leaf. The humidi ed box was placed in an 18 °C incubator with light for humidi ed culture (8 h in darkness and 16 h in light every day). The disease condition was observed 5 days after inoculation, and data for statistical analysis were collected. Ten leaves were inoculated for each treatment.

Indoor investigation methods
Five days after inoculation, the different disease severity levels were determined according to the diameter of the lesion, and the diameter of the lesion was measured by the cross method. The diameter of the lesion = 1/2 × (length + width) (unit: mm). The speci c standards are shown in Table 6. The diameter of the lesion: 5 mm d ≤ 10 mm; the lesion is water immersed with chlorotic halo on the edge; uniform mold layer is produced on the lesion, but the mold layer is sparse.
Moderately Susceptible 3 The diameter of the lesion: 10 mm d ≤ 20 mm; the lesion was water soaked, with chlorotic halo on the edge; uniform mold layer was produced on the lesion, but the mold layer was thick.
Highly susceptible 4 The diameter of the lesion: d > 20 mm; uniform and dense mould layer appeared on the lesion.

DNA preparation and sequencing
The improved cetyl trimethylammonium bromide (CTAB) method [37] was used to extract genomic DNA from potato leaves. After detection and quanti cation of the concentration by 1% agarose gel electrophoresis, the working DNA solution was diluted to 100 ng/µl and stored at -20 °C. The ampli ed-fragment SNP and methylation (AFSM) approach [38] was then used to construct EcoRI-MspI and EcoRI-HpaII libraries of the 284 potato DNA samples. After the monoclonal detection met the requirements, the EcoRI-MspI and EcoRI-HpaII libraries were mixed into one library at a ratio of 1:1, and HiSeq 2500 was used to perform paired-end 150-bp sequencing on the constructed sequencing library.

SNP calling and annotation
We used a Perl script (http://afsmseq.sourceforge.net/) to lter the original sequencing data, count the total number of reads obtained from sequencing, assign the reads to each individual based on the barcodes designed using the AFSM technology, and count the number of reads in each individual. Bowtie2 software [39] was used to align the optimized sequencing reads to the potato DM reference genome (http://solanaceae.plantbiology.msu.edu/), and SAMtools [40] and VCFtools (http://vcftools.sourceforge.net/) were used to detect SNP and indel loci. Based on the potato DM reference genome v4.03, the snpEff software [41] was used to identify the mutation locations (intergenic region, untranslated region/UTR, gene upstream region, or gene downstream region), mutation types (synonymous, missense, frameshift, and non-frameshift), and annotate them at the same time.
Analysis of population structure and genetic diversity We rst used PHYLIP (http://evolution.genetics.washington.edu/phylip.html) to calculate the genetic distance matrix of the sample. We then used the Notepad + + software to save the genetic distance matrix le in a suitable format. A phylogenetic tree was constructed using the neighbour joining method. After generating the tree le, iTOL (https://itol.embl.de/) was used to draw the phylogenetic tree diagram. GCTA software was used to conduct principal-component analysis (PCA) of the potato population materials with the detected SNPs as inputs [42]. R software was then used to calculate the vector of each principal component and draw the PCA scatter plot. In addition, ADMIXTURE software [43] was used to analyse the population structure and estimate the optimal number of population subgroups. PLINK software [44] was used to adjust the input le format for ADMIXTURE software, and then we input the le. The K value range of the number of subgroups was set to 1-12, and the appropriate value of K for the number of subgroups was determined according to the obtained crossvalidation error value. The genetic composition coe cient (Q) of each material in each subgroup was used to construct the population-genetic structure matrix. VCFtools software (https://vcftools.github.io/index.html) was used to calculate the genetic diversity (π) and population pairwise F-statistics (F ST ) [45].

LD analysis
In the entire population and each subgroup (inferred using ADMIXTURE), the value of r 2 was used to evaluate the LD relationship between each pair of polymorphic sites throughout the genome, and the value of r 2 was calculated by PopLDdecay software [47] for high-quality SNPs after ltering. The genetic distance was sorted from small to large, and then the average value of LD r 2 in the segment was calculated to draw a scatter lot with a smooth curve. The genetic distance interval where the curve intersected with the straight line representing a non-collinear r 2 of 99% was the LD decay distance.

Association analysis
In this study, data from 22,489 high-quality SNPs and indels were typed to perform a genome-wide association analysis on the severity (lesion diameter) and resistance grade of this population. We used the compressed mixed linear model of TASSEL 5.0 software [48] for correlation analysis. The kinship matrix was obtained with the same software, and the threshold for the signi cance of the severity was set at 1/n, where n represents the number of all SNPs. The threshold for the signi cance of resistance grade was set at 0.05. Inputting the physical location of the SNP in the potato genome and its P value, the qqman package of R software was used to draw the Manhattan plot. SAMtools was used to manually verify regions that were signi cantly correlated with the reordered read results of the potato reference genome PGSC_DM_v4.03.

Candidate gene screening
Based on potato SNP annotation and LD decay, and according to the functional annotation of the loci, the genes where the loci were located were used as the candidate genes. If a locus was located upstream and downstream of other genes at the same time, the upstream and downstream genes were also used as candidate genes. If a locus was located in the intergenic region, the upstream and downstream genes that were closest to the locus were used as candidate genes. A mixed linear model was used to perform association analysis on the traits of late blight.
qRT-PCR veri cation Primers were designed based on the coding region sequence (CDS) of the candidate genes, and the actin gene was the internal reference (Table 7). A uorescence reverse transcription kit (TaKaRa) was used to generate cDNA using 500 ng RNA as the template. A uorescence quantitative RT-PCR kit (TaKaRa) was used to perform quantitative real-time polymerase chain reaction (qRT-PCR) with the thermal cycling programme of 95 °C 30 s followed by 40 cycles of 95 °C 5 s and 60 °C 30 s. Excel 2016 was used to sort and analyse the gene expression uorescence qRT-PCR (Quantitative real-time polymerase chain reaction) data, and the 2 −ΔΔCt method was used to calculate the relative expression levels. Table 7 qRT-PCR primers for the selected genes