Dataset Information

A Deep Learning Approach to Population Structure Inference in Inbred Lines of Maize.

ABSTRACT: Analysis of population genetic variation and structure is a common practice for genome-wide studies, including association mapping, ecology, and evolution studies in several crop species. In this study, machine learning (ML) clustering methods, K-means (KM), and hierarchical clustering (HC), in combination with non-linear and linear dimensionality reduction techniques, deep autoencoder (DeepAE) and principal component analysis (PCA), were used to infer population structure and individual assignment of maize inbred lines, i.e., dent field corn (n = 97) and popcorn (n = 86). The results revealed that the HC method in combination with DeepAE-based data preprocessing (DeepAE-HC) was the most effective method to assign individuals to clusters (with 96% of correct individual assignments), whereas DeepAE-KM, PCA-HC, and PCA-KM were assigned correctly 92, 89, and 81% of the lines, respectively. These findings were consistent with both Silhouette Coefficient (SC) and Davies-Bouldin validation indexes. Notably, DeepAE-HC also had better accuracy than the Bayesian clustering method implemented in InStruct. The results of this study showed that deep learning (DL)-based dimensional reduction combined with ML clustering methods is a useful tool to determine genetically differentiated groups and to assign individuals into subpopulations in genome-wide studies without having to consider previous genetic assumptions.

SUBMITTER: Lopez-Cortes XA

PROVIDER: S-EPMC7732446 | biostudies-literature | 2020

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

A Deep Learning Approach to Population Structure Inference in Inbred Lines of Maize.

López-Cortés Xaviera Alejandra XA Matamala Felipe F Maldonado Carlos C Mora-Poblete Freddy F Scapim Carlos Alberto CA

Frontiers in genetics 20201124

Analysis of population genetic variation and structure is a common practice for genome-wide studies, including association mapping, ecology, and evolution studies in several crop species. In this study, machine learning (ML) clustering methods, K-means (KM), and hierarchical clustering (HC), in combination with non-linear and linear dimensionality reduction techniques, deep autoencoder (DeepAE) and principal component analysis (PCA), were used to infer population structure and individual assignm ...[more]

PMID: 33329691

Similar Datasets

Project description:BackgroundMolecular characterization is important for efficient utilization of germplasm and development of improved varieties. In the present study, we investigated the genetic purity, relatedness and population structure of 265 maize inbred lines from the Ethiopian Institute of Agricultural Research (EIAR), the International Maize and Wheat Improvement Centre (CIMMYT) and the International Institute of Tropical Agriculture (IITA) using 220,878 single nucleotide polymorphic (SNP) markers obtained using genotyping by sequencing (GBS).ResultsOnly 22% of the inbred lines were considered pure with <5% heterogeneity, while the remaining 78% of the inbred lines had a heterogeneity ranging from 5.1 to 31.5%. Pairwise genetic distances among the 265 inbred lines varied from 0.011 to 0.345, with 89% of the pairs falling between 0.301 and 0.345. Only <1% of the pairs had a genetic distance lower than 0.200, which included 14 pairs of sister lines that were nearly identical. Relative kinship analysis showed that the kinship coefficients for 59% of the pairs of lines was close to zero, which agrees with the genetic distance estimates. Principal coordinate analysis, discriminant analysis of principal components (DAPC) and the model-based population structure analysis consistently suggested the presence of three groups, which generally agreed with pedigree information (genetic background). Although not distinct enough, the SNP markers showed some level of separation between the two CIMMYT heterotic groups A and B established based on pedigree and combining ability information.ConclusionsThe high level of heterogeneity detected in most of the inbred lines suggested the requirement for purification or further inbreeding except those deliberately maintained at early inbreeding level. The genetic distance and relative kinship analysis clearly indicated the uniqueness of most of the inbred lines in the maize germplasm available for breeders in the mid-altitude maize breeding program of Ethiopia. Results from the present study facilitate the maize breeding work in Ethiopia and germplasm exchange among breeding programs in Africa. We suggest the incorporation of high density molecular marker information in future heterotic group assignments.

Project description:Maize (Zea mays L.) is an important cereal and is affected by climate change. Therefore, the production of climate-smart maize is urgently needed by preserving diverse genetic backgrounds through the exploration of their genetic diversity. To achieve this, 96 maize inbred lines were used to screen for phenotypic yield-associated traits and grain quality parameters. These traits were studied across two different environments (Anand and Godhra) and polymorphic simple sequence repeat (SSR) markers were employed to investigate the genetic diversity, population structure, and trait-linked association. Genotype-environment interaction (GEI) reveals that most of the phenotypic traits were governed by the genotype itself across the environments, except for plant and ear height, which largely interact with the environment. The genotypic correlation was found to be positive and significant among protein, lysine and tryptophan content. Similarly, yield-attributing traits like ear girth, kernel rows ear-1, kernels row-1 and number of kernels ear-1 were strongly correlated to each other. Pair-wise genetic distance ranged from 0.0983 (1820194/T1 and 1820192/4-20) to 0.7377 (IGI-1101 and 1820168/T1). The SSRs can discriminate the maize population into three distinct groups and shortlisted two genotypes (IGI-1101 and 1820168/T1) as highly diverse lines. Out of the studied 136 SSRs, 61 were polymorphic to amplify a total of 131 alleles (2-3 per loci) with 0.46 average gene diversity. The Polymorphism Information Content (PIC) ranged from 0.24 (umc1578) to 0.58 (umc2252). Similarly, population structure analysis revealed three distinct groups with 19.79% admixture among the genotypes. Genome-wide scanning through a mixed linear model identifies the stable association of the markers umc2038, umc2050 and umc2296 with protein, umc2296 and umc2252 with tryptophan, and umc1535 and umc1303 with total soluble sugar. The obtained maize lines and SSRs can be utilized in future maize breeding programs in relation to other trait characterizations, developments, and subsequent molecular breeding performances for trait introgression into elite genotypes.

Dataset Information

A Deep Learning Approach to Population Structure Inference in Inbred Lines of Maize.

Publications

A Deep Learning Approach to Population Structure Inference in Inbred Lines of Maize.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets