The Development of Quality Control Genotyping Approaches: A Case Study Using Elite Maize Lines

Abstract

Quality control (QC) of germplasm identity and purity is a critical component of breeding and conservation activities. SNP genotyping technologies and increased availability of markers provide the opportunity to employ genotyping as a low-cost and robust component of this QC. In the public sector available low-cost SNP QC genotyping methods have been developed from a very limited panel of markers of 1,000 to 1,500 markers without broad selection of the most informative SNPs. Selection of optimal SNPs and definition of appropriate germplasm sampling in addition to platform section impact on logistical and resource-use considerations for breeding and conservation applications when mainstreaming QC. In order to address these issues, we evaluated the selection and use of SNPs for QC applications from large DArTSeq data sets generated from CIMMYT maize inbred lines (CMLs). Two QC genotyping strategies were developed, the first is a "rapid QC", employing a small number of SNPs to identify potential mislabeling of seed packages or plots, the second is a "broad QC", employing a larger number of SNP, used to identify each germplasm entry and to measure heterogeneity. The optimal marker selection strategies combined the selection of markers with high minor allele frequency, sampling of clustered SNP in proportion to marker cluster distance and selecting markers that maintain a uniform genomic distribution. The rapid and broad QC SNP panels selected using this approach were further validated using blind test assessments of related re-generation samples. The influence of sampling within each line was evaluated. Sampling 192 individuals would result in close to 100% possibility of detecting a 5% contamination in the entry, and approximately a 98% probability to detect a 2% contamination of the line. These results provide a framework for the establishment of QC genotyping. A comparison of financial and time costs for use of these approaches across different platforms is discussed providing a framework for institutions involved in maize conservation and breeding to assess the resource use effectiveness of QC genotyping. Application of these research findings, in combination with existing QC approaches, will ensure the regeneration, distribution and use in breeding of true to type inbred germplasm. These findings also provide an effective approach to optimize SNP selection for QC genotyping in other species.

Fig 1. Summary of the heterogeneity, minor allele frequency (MAF) and polymorphic information content (PIC) of 18,082 selected SNPs.

Chromosome assignments are indicated; where no BLAST position was available, the chromosome is designated as “0”; The heterogeneity, MAF, percentage of Missing value, PIC was shown in left y-axis, the number of marker for each chromosome was shown in right y-axis.

Fig 2. Analysis of methods to enable selection of markers for QC genotyping.

a: MAF effect. Random selection (Random); MAF < 0.15 (0.05–0.15); MAF between 0.15 and 0.25 (0.15–0.25); MAF > 0.25 (>0.25); b: Marker Group effect on select marker for QC. Random selection (Random); Same percentage of marker from each marker group (PG); same number of markers selected from each group (NG); Keep the proportion of each group distance (PGD). c: Marker coverage effect. Random selection (Random); Coverage < 2 (0–2); Coverage between 2 and 15 (2–15); Coverage >15 (>15). d: Marker distribution effect. Standard error bars are shown.

Fig 3. Association analysis indentifies QPM and imidazolinone resistance markers.

Fig 4. Comparison of the effect of the final marker selection rules versus random marker selection on the proportion of CML pairs not distinguished from one another.

Fig 5. Comparison of five subsets of SNPs for broad QC and rapid QC.

a: Five subsets of 80 SNPs for broad QC genotyping; b: Five subsets of 10 SNPs for rapid QC genotyping.

Fig 6. Probability of detecting at least one off-type using different sample sizes at different assumed off-type levels (P) within populations.