genetic markers & Models

GWAS & Genomic Selection

In Chawade lab we develop genomics tools for plant breeding using linkage mapping, GWAS and genomic selection.

Linkage mapping

Linkage mapping identifies genetic markers linked to traits by analyzing inheritance patterns. 


Genome-wide association studies (GWAS) detect associations between traits and markers across the genome to identify candidate genes. 

Genomic Selection

Genomic selection predicts breeding values based on marker profiles, aiding in the selection of individuals with desirable traits for breeding programs.

Linkage Mapping & GWAS

In Chawade lab we identified several major qtl for important diseases in winter wheat, barley, potato and rape seed. Linkage mapping and GWAS are used as appropriate. Please visit google scholar for the list of publications utilizing these methods

Genomic Selection

Chawade lab explores various methods to enhance the accuracy of genomic selection (GS) by employing multiple GS algorithms, selecting the number of SNPs, considering genotype-by-environment interactions, and using GWAS as covariates. We’ve demonstrated that just 200-2000 SNPs, identified through haplotyping and bioinformatics, are adequate for genomic prediction in wheat, potato and barley. Currently, we are integrating GS into the conventional breeding program for pea at Findus, winter wheat and spring barley at Lantmännen

Key findings from our work in genomic selection:

Integrating fixed effects improves accuracy of genomic prediction

Prediction accuracy increases with fixed effects:

  • From 0.47 to 0.62 for STB in genbank germplasm (Odilbekov et al. 2019).
  • From 0.32 to 0.61 for STB and from 0.33 to 0.83 for powdery mildew in genebank germplasm (Alemu et al. 2021)
  • For multi-environment genomic prediction (Alemu et al. 2021)
  • From 0.49 to 0.58 for STB in breeding germplasm (Zakieh et al. 2023)
Decreasing genotyping costs using fewer SNPs
  • Fewer than 1000 SNPs (based on LD thresholding) produces models with equally good prediction accuracy.
  • As few as 300 SNPs with marginal decrease in prediction accuracies in winter wheat for FHB. (Alemu et al. 2023)
  • A genotyping panel with just 200 SNPs developed for barley and now being used in the barley breeding program.

Training population optimization is necessary

  • Poor prediction accuracies when training and test populations differ significantly
  • In winter wheat accuracies are low (<0.2) when training on genebank germplasm and testing on breeding germplasm and vice versa
  • In barley, accuracies are low (<0.2) when training on six-row barely and testing on two-row barley and vice versa (Åstrand et al., under review)
  • In faba bean, there is a partial overlap between selected genebank and breeding germplasm, leading to anticipated challenges.
  • Population structure issues can not be resolved using different GS algorithms