AUTHOR=Yang Yang , Tian Hongli , Wang Rui , Wang Lu , Yi Hongmei , Liu Yawei , Xu Liwen , Fan Yaming , Zhao Jiuran , Wang Fengge 

TITLE=Variety Discrimination Power: An Appraisal Index for Loci Combination Screening Applied to Plant Variety Discrimination

JOURNAL=Frontiers in Plant Science

VOLUME=Volume 12 - 2021

YEAR=2021

URL=https://www.frontiersin.org/journals/plant-science/articles/10.3389/fpls.2021.566796

DOI=10.3389/fpls.2021.566796

ISSN=1664-462X

ABSTRACT=Molecular marker technology is used widely in plant variety discrimination, molecular breeding, and in other fields. To lower the cost of testing and improve the efficiency of data analysis, molecular marker screening is very important. Screening usually involves two phases, the first to control loci quality and the second to reduce loci quantity. To reduce loci quantity, an appraisal index that is very sensitive to a specific scenario is necessary to select loci combinations. In this study, we focused on loci combination screening for plant variety discrimination. A loci combination appraisal index, variety discrimination power (VDP), is proposed, and three statistic methods, probability-based VDP (P-VDP), comparison-based VDP (C-VDP), and ratio-based VDP (R-VDP), are described and compared. The results using the simulated data showed that VDP was sensitive to statistical populations with convergence toward the same variety and the total probability of discrimination power (TDP) method was effective only for partial populations. R-VDP was more sensitive to statistical populations with convergence toward various varieties than P-VDP and C-VDP, which both had the same sensitivity; TDP was not sensitive at all. With the real data, the differences between estimated R-VDP values of 20-loci combinations and 1 were in orders of magnitude from E−2 to E−1, between P-VDP or C-VDP values and 1 they were from E−4 to E−3, and between TDP values and 1 they were from E−9 to E−8. For the variety threshold setting, R-VDP values of loci combinations with different numbers of loci responded evenly to different thresholds. C-VDP values responded unevenly to different thresholds and the extent of the response increased as the number of loci decreased. All the methods gave underestimations when data were missing, with systematic errors for TDP, C-VDP, and R-VDP going from biggest to smallest. We concluded that VDP was a better loci combination appraisal index than TDP for plant variety discrimination and the three VDP methods have different applications. We developed the software called VDPtools, which can calculate the values of TDP, P-VDP, C-VDP and R-VDP. VDPtools is publicly available at: https://github.com/caurwx1/VDPtools.git