An article titled “A graph-based genome and pan-genome variation of the model plant Setaria” [1], published in Nature Genetics, de novo assembled 110 reference-grade genomes for cultivated Setaria accessions, and examined genome evolution in the context of foxtail millet domestication and improvement.
Setaria italica (foxtail millet) is an ideal model for C4 photosynthetic crop plants. Despite its favorable traits, genomic diversity and potential for genetic enhancement remain relatively understudied. Through high throughput sequencing, 110 core-set reference-level genomes were assembled, facilitating the establishment of a complete pan-genome and the construction of a graph-based genome for Setaria. Large-scale genetic studies across 68 traits in 13 environments identified potential genes for millet improvement in diverse geographic locations. These findings can inform marker-assisted breeding, genomic selection, and genome editing to expedite crop improvement under various climatic conditions.
Background
Foxtail millet (Setaria italica), as one of the oldest domesticated grain crops in the world, has excellent drought and low soil-nutrient tolerance thriving in temperate, tropical, and arid environments. Significantly, Setaria species employ C4 photosynthesis, characterized by high photosynthetic efficiency and environmental adaptability, crucial for global agricultural grain and biofuel production. Because of the complexity of most C4 crop plant genomes and the limited high-efficiency transformation systems, Setaria makes it an attractive model for the study characterized with compact diploid genomes (~420 Mb), short life cycles (~70 d), and efficient transformation systems. The study can elucidate genotyping for crop domestication and genetic improvement, paving the way for foxtail millet research and breeding, and serving as a model for 'breeding by design' in other crops.
Study Design and Methods
The genomic resequencing data for 630 wild, 829 landrace, and 385 modern cultivated accessions from the Setaria genus were collected by the researchers, with an average sequencing depth of approximately 15×. After aligning reads to the foxtail millet ‘Yugu1’ reference genome, ~60 million SNPs and 6.7 million InDels were identified. Phylogenetic and population structure analyses were conducted to reveal genetically differentiated subpopulations within wild accessions (W1-W4) and cultivated accessions (C1–C3).
To capture the complete spectrum of genetic diversity of Setaria which may be overlooked by short-read resequencing approaches, 110 representative Setaria accessions (35 wild, 40 landrace and 35 modern cultivated accessions) were assembled and assessed by K-mer-based analysis. PacBio and Illumina reads were used to further refine three representative genomes.
A pan-genome of foxtail millet was constructed, and SV distribution analysis was performed. Phylogenetic analysis incorporating SVs and Presence-absence variants (PAV) frequency comparisons was conducted to elucidate domestication associations. Genome-wide selection signatures for domestication were identified using SNP data.
To identify seed-shattering loci, they performed quantitative trait loci (QTL) analysis and bulked segregant analysis sequencing (BSA-seq) using an RIL population. Through this approach, three major QTLs associated with seed shattering were identified. The SV-based GWAS was conducted to elucidate the relationship between SV and grain yield increase. Subsequently, overexpression experiments in foxtail millet were performed to validate the functional role of candidate genes. A graph-based reference genome of Setaria was constructed by integrating insertions, deletions and inversions across 112 accessions into the reference genome. This resource accounts for pan-genome variation and provides a valuable tool for breeding programs.
Results and Discussion
Evolution and De novo Assembly of Setaria: According to phylogenetic and population structure analyses using 4,934,413 high-quality SNPs, W1 subgroup emerges as the closest population to cultivated foxtail millet, suggesting it as the wild progenitor for all cultivated accessions. Among cultivated subgroups, the broad distribution of C3 subgroup worldwide implies potential adaptation to a wider range of climates. 110 representative Setaria accessions were de novo assembled and three accessions—Me34V (wild), Ci846 (landrace) and Yugu18 (modern cultivar)—were further selected to construct high-quality reference genome assemblies for Setaria. The three genome assemblies have greater contiguity than existing reference genomes, with a mean contig N50 length of >20 Mb and LTR assembly index >20.
Pan-genomic Variation in Foxtail Millet Domestication and Improvement: Comprising over 100 Setaria accessions, the pan-genome suggested 23.8% were core genes, 42.9% were soft core genes, 29.4% were dispensable and 3.9% were private genes. Most PAVs were found to overlap with transposable elements (TEs), indicating TE activity as an important mechanism for SV generation in genomes. A set of SVs was identified within promoters or gene bodies of functionally important loci, and SVs occur more frequently in genes with low expression level. This pattern mirrors observations in rice [2] and aligns with a stabilizing model of gene expression evolution, wherein lowly expressed genes are under weaker selection pressure and thus more prone association with PAVs. Similar to the studies of other crops, SVs play a crucial role in determining foxtail millet traits. This is exemplified by the study of two key domestication genes, SiGW3 and sh1.
Graph-based Genome Facilitates Breeding of Foxtail Millet: Using the graph-based genome, researchers genotyped SVs in a large population via short-read resequencing. Subsequently, they conducted GWAS and genomic selection (GS) studies in 680 foxtail millet accessions across 68 traits in 13 distinct geographic locations with varying climatic conditions. SNPs and SVs significantly associated with diverse phenotypes were identified, enhancing genomic prediction accuracy for foxtail millet across different environments. This prediction accuracy is substantially higher than observed in tomato12 possibly due to species or trait specificity. Leveraging the graph-based genome, the potential breeding values for yield and grain quality-related traits can be estimated, offering avenues for foxtail millet breeding strategies tailored to climate change adaptation.
Reference [1] He, Q., Tang, S., Zhi, H. et al. A graph-based genome and pan-genome variation of the model plant Setaria. Nat Genet 55, 1232–1242 (2023). [2] Qin, Peng et al. “Pan-genome analysis of 33 genetically diverse rice accessions reveals hidden genomic variations.” Cell vol. 184,13 (2021).
this type of routine de novo assembly of genomes is awesome, great work