There Is a Secret to Grape Flavor. The Academy of Agricultural Sciences Uses Machine Learning to Reveal the Process of Gene Introgression

Contents at a glance:Genetic introgression is closely related to the domestication and genetic improvement of grapes. Previous studies have revealed genomic signals of wild grape gene introgression in European cultivated grapes, but the timing, mode, genomic pattern and biological effects of these introgression events have not been studied in depth. In this paper, researchers from the Shenzhen Institute of Agricultural Genomics, Chinese Academy of Agricultural Sciences, used population genetic analysis methods based on machine learning to analyze resequencing data of cultivated and wild grapes, which is of great significance for grape breeding.

Keywords:Genetic introgression Grape domestication Machine learning

This article was first published on HyperAI WeChat public platform~

Gene introgression refers to the process by which genes are introduced from one species or population into another.It usually occurs during hybridization and backcrossing between different species. In gene introgression, foreign genes enter the gene pool of the target species through hybridization and mix with the native genes of the target species.This introgression can lead to changes in the genome of the target species, acquiring new genetic variation and diversity.

Grape introgression refers to the process of gene exchange and gene flow between wild grapes (Vitis vinifera ssp. sylvestris) and cultivated grapes (V. vinifera ssp. vinifera).Under the influence of nature and artificial selection, gene exchange occurs between wild grapes and cultivated grapes, resulting in the appearance of wild grape genetic characteristics in cultivated grapes.

Recently, researchers from the Shenzhen Agricultural Genomics Center of the Chinese Academy of Agricultural Sciences studied the introgression history between domesticated grapes and their wild European relatives. Using a machine learning-based population genetics approach, they revealed the mechanism of grape flavor formation and provided in-depth insights into its genetic characteristics and its impact on grape breeding.Currently, the research results have been published in the journal Proceedings of the National Academy of Sciences, titled "Adaptive and maladaptive introgression in grapevine domestication."

The research results have been published in the journal Proceedings of the National Academy of Sciences

Paper address:

https://www.pnas.org/doi/abs/10.1073/pnas.2222041120

Dataset

The researchers analyzed a set of 345 sequenced samples, including wine grapes and table grapes, covering the broad geographic distribution range of wild grapes.These included 72 wild grape species (V. vinifera ssp. sylvestris) from Europe, 36 wild species from the Middle East and the Caucasus, and 231 domesticated grape species (V. vinifera ssp. vinifera), as well as North American outgroup species Vitis californica (n = 3) and Muscadinia rotundifolia (n = 3). In addition, the researchers sampled European wild grapes to cover their predicted distribution areas in Europe and the Near East.

The data (genome sequence; script) has been deposited in GitHub:

https://github.com/zhouyflab/Grapevine_Adaptive_Maladaptive_Introgression

Experimental Results

History of introgression 

First, the researchers assessed the differences between European wild grapes, wine grapes, and table grapes to gain insight into population and domestication history.Notably, wild European grapes (EU sylvestris) form a distinct group, sharing few admixture components with wine grapes.

Figure 1. Comparison of different grape varieties

A:Phylogenetic tree of the mixed analysis.

In the phylogeny, the colors of the branches reflect the different groups: ME 1, yellow; ME 2, purple; wild grapes, reddish-brown; wine grapes, blue; table grapes, green. Admixture plot, K = 6. The red dots and blue triangles on the right side of the plot show whether chloroplasts or mitochondria from the food or wine group, respectively, clearly originated from European wild grapes.

B:PCA of the five groups.

C:Heterozygosity of the five groups.

D:Forward simulation results for different propagation types. The blue line represents outcrossing, while the orange line represents clonal propagation.

Figure 1 shows that wine grapes and table grapes diverged early in their evolution, indicating that they have distinct uses. Wild grape samples also showed a monophyletic group, but were divided into three different groups: European wild grapes (EU), grapes around the Caspian Sea (ME 1), and grapes in the fertile crescent near the Mediterranean Sea (ME 2). The differentiation between these populations was also confirmed by principal component analysis (PCA) and estimation of the proportion of ancestral components.The heterozygosity of wine grapes and table grapes (both 0.24) was higher than that of wild populations (0.17 in Europe, 0.20 in ME1, and 0.22 in ME2), which may be due to the accumulation of heterozygous mutations caused by historical introgression events and long-term asexual reproduction.

Direction of gene introgression 

Building on their initial model, the researchers estimated 34 possible patterns of gene flow between table grapes, wine grapes and European wild grapes.Based on the best model, fastsimcoal inferred that European wild grapes diverged around 40,000 years ago.

Figure 2: Introgression assay

The time (T) in the horizontal branches is the divergence time of each group (years), and Ti represents the inferred introgression start time.

The numbers at the bottom are the estimated effective population sizes (Ne) for each group.

Figure 2 shows that under the best model, fastsimcoal inferred that the EU group diverged about 4 × 104 years ago; the domestication of table grapes began as early as 1.5 × 104 years ago; and wine grapes separated from table grapes about 1.0 × 104 years ago. The model shows that gene flow between wild and domesticated populations in Europe began 1.8 × 103 years ago.In addition, the best-fit model also showed that the probability of gene flow from European wild populations to domesticated populations was high, with the migration rate to wine grapes (1.7 × 10−4) being five times that to table grapes (3.8 × 10−5).

Introgression region 

The researchers used machine learning methods to identify introgressed regions of the grapevine genome, as well as the genomic features of these regions.

Figure 3:Three genes in the introgressed region

The putative introgression regions predicted by Filet are marked with black lines on the 19 chromosomes.

The three colors represent the genes in the three introgression regions.

red:Flowering related genes

green:Aroma compound related genes

blue:Stress response genes.

The researchers conducted a GO functional study of introgression selection genes across the entire genome.And focus on flowering-related genes, flavor-related genes and stress response-related genes.Figure 3 shows that many flowering-related genes were selected after introgression, followed by aromatic compound-related genes. Most GO categories enriched were also related to flavor, including lignin degradation process, L-phenylalanine degradation process, and cinnamic acid biosynthesis process.Therefore, the researchers infer that adaptive introgression between wine grapes and European wild grapes primarily affected flavor-related traits.

To understand the introgression dynamics of alleles,The researchers performed forward simulations of introgression from outbreeding to clonal reproduction.

Figure 4:SFS of beneficial and deleterious SNPs in non-introgressed and introgressed regions

E:The total number of introgressed alleles in the entire introgressed population

F:The number of different types of introgressed alleles in the entire introgressed population

G:SFS of introgressed beneficial and deleterious alleles in the 500th generation of outcrossing group

H:SFS of beneficial and deleterious alleles introgressed in the 500th generation of the clone group

The study found that the total number of introgressed alleles (both beneficial and deleterious alleles) increased after hybridization until equilibrium was reached. These simulations suggest that the profile of introgressed alleles may differ significantly between clonal and outcrossing systems. Introgression plays an important role in the recruitment of beneficial and deleterious variation, which may be a major target for genomic design in grapevine cultivation, including the elimination of potentially deleterious variation during sexual reproduction.

AGIS: Focus on agricultural technology innovation

It is worth noting that many authors of this paper are from the Shenzhen Institute of Agricultural Genomics, Chinese Academy of Agricultural Sciences.The Institute of Genomics was established in 2014. It integrates biology and big data science to understand and utilize agricultural biological genomes to serve global agricultural production.The long-term vision of the Institute of Genomics is to promote sustainable global agriculture through disruptive innovation, serve a personalized food supply system, and improve human health and the social status of farmers.

The Institute of Genomics has published more than 620 SCI papers in top journals including Science, Nature, Cell, and is at the forefront of the world in research areas such as agricultural genomics.In addition to studying the problem of gene introgression in grapes, the institute also published two other research results related to grape genes. It not only collaborated with scholars at home and abroad to draw the complete genome map of grapes for the first time, but also revealed the genome-wide effects and climate adaptability mechanisms of wild grapes' resistance to Pierce's disease.

The director said that the Institute of Genomics is still a very young research institute, and there is a long way to go to build a world-class agricultural research institute. We will stand on the new era, new journey and new mission, adhere to the "four aspects" to help high-level scientific and technological self-reliance, and continue to make our contribution to building a "strong agricultural country". At present, the Institute of Genomics has jointly proposed the construction of the "Shenzhen International Food Valley" with relevant departments of Shenzhen City, and the plan has been approved by the municipal government.We will build an agricultural and food industry-university-research collaboration ecosystem in Shenzhen and provide a pioneering example of how science and technology can drive the transformation and upgrading of the agricultural and food industry.

Genomics Institute official website:

https://www.agis.org.cn/index.htm

References:

[1]https://www.163.com/dy/article/I6KVQLV205328VPM.html

[2]https://www.sohu.com/a/682674856_121124027

[3]https://www.caas.cn/xwzx/kyhd/60f2e9b4dff84bed9e315b7097aeb26b.htm

[4]https://www.agis.org.cn/bsgk/yjsjj/index.htm

[5]https://www.agis.org.cn/xwzx/kyjz/677aecae97c448c9bed7e89f95daae7f.htm