Marker assisted selection (MAS) is an effective selection strategy that has potential to augment conventional phenotype-based selection. Precision of marker assisted selection largely depends on identifying markers closely linked to genomic regions governing trait of interest. Under this premise, Association mapping (AM) via Genome wide association studies (GWAS) emerged as a potential tool to identify markers linked to traits with complex inheritance, especially in the crops where mapping populations are difficult to synthesise.

AM relies on the principle of linkage disequilibrium (LD), which is defined as non-random association of alleles in the population. Various means of quantifying LD have been reported in studies, of which r2 is commonly used, due to its ability to account for both recombinational and mutational history3. Natural populations have been the choice for AM, owing to their large availability in crop species especially where hybridization is cumbersome to develop bi/multiparental mapping populations. Depending on the constitution of mapping panel, several statistical issues like population structure, kinship, rare alleles, false discovery rate could largely influence the detected associations, resulting in increased false positives and false negatives. Hence, it becomes imperative to address these concerns statistically, to identify markers truly associated with traits of interest.

To address the confounding effects in detecting marker trait associations, single locus and multi loci statistical models like GLM, MLM, CMLM, SUPER, FARMCPU, BLINK etc., have been developed, that include population structure and (or) kinship as covariates in the multiple regression model, to reduce type I and type II errors. Comparison of these models in simulated datasets of Soyabean, with varying number of QTLs and trait heritability revealed FARMCPU as the best model1. Above-mentioned models detect independent SNPs associated with target traits, ignoring the possible interactions among them and with the environment. This is circumvented with the deployment of compressed variance component mixed linear models (3VmrMLM), which detect causal Quantitative trait Nucleotides (QTNs), QTN-QTN and QTN and environment interactions2. Thereafter, candidate genes underlying these detected QTNs, which are further screened to identify haplotypes governing the trait. This approach identified superior haplotype combinations in the vicinity of QTNs controlling seedling salinity tolerance in Rice4.

With advancement in programming, a plethora of models to detect marker trait associations have been developed and are being successfully utilised. These models complement each other in demarcating true and causal associations from false positives, which after substantial validation in other mapping populations, could be deployed in marker assisted selection for achieving rapid genetic gain.

References: 

1. KALER, A. S., GILLMAN, J. D., BEISSINGER, T. AND PURCELL, L. C., 2020, Comparing different Statistical models and multiple testing corrections for association mapping in Soybean and Maize. Front. Plant Sci., 10:1794. 

2. LI, M., ZHANG, Y. W., ZHANG, Z. C., XIANG, Y., LIU, M. H., ZHOU, Y. H., ZUO, J. F., ZHANG, H. Q., CHEN, Y., AND ZHANG, Y. M., 2022, A compressed variance component mixed model for detecting QTNs and QTN-by-environment and QTN-by-QTN interactions in genome-wide association studies. Mol. Plant, 15: 630–650. 

3. SINGH, B.D. AND SINGH, A.K., 2015, Marker-assisted plant breeding: principles and practices. Springer Nature, pp: 217-255.

4. ZHANG, G., BI, Z., JIANG, J., LU, J., LI, K., BAI, D., WANG, X., ZHAO, X., LI, M., ZHAO, X., WANG, W., XU, J., LI, Z., ZHANG, F. AND SHI, Y., 2023, Genome-wide association and epistasis studies reveal the genetic basis of saline-alkali tolerance at the germination stage in rice. Front. Plant Sci., 14:1170641.