Statistical Tools for Seed Testing
A collection of statistical programs and modules to assist ISTA members in their routine work and method development and validation.
These programs were made for educational purposes. You may download these files and use them freely. ISTA and the authors shall not be liable for any loss, damage, etc. resulting from their use.
Interlaboratory tests using ISO 5725 - Sylvain Gregoire (FR)
The Zip file contains an Access97© and an Access 2000© database, a Powerpoint© presentation, a short help file, an example of the file containing data in appropriate format to include in the database. By use of requests on the data from an experiment 3 types of reports are provided (screen, paper, HTML) as examples of what can be done. Descriptive statistics and graphs are provided. Repeatability and reproducibility values as described in ISO 5725-2 are computed, and h and k tests identify deviating labs. This can be used when different labs test lots of seeds with the same method, and replicates of quantitative measurements are available for each lot in each laboratory. The piece of software can be modified by the user, but inappropriate actions can cause the computations to become false.
Statistical tests for seed mixtures - Sylvain Gregoire (FR), Julia Barabas (HU)
A Powerpoint© presentation introduces the work which resulted in an Excel© file with four types of Chi2 tests for seed mixtures. One can use buttons to navigate through the spreadsheet or use Excel© classically. Data can be directly typed into the cells. Cells with formulae are protected, but the user is able to remove the sheet protection in order to adapt the tool to his needs.
Seedcalc8 - Kirk Remund (US), Robert Simpson (US), Jean-Louis Laffont (FR), Deanne Wright (US), Sylvain Gregoire (FR)
Seedcalc8 is a Microsoft Excel® application that is written for Windows 2000 and XP. It can be used to design seed testing plans for purity/impurity characteristics including testing for adventitious presence levels of biotech traits in conventional seed lots. This application can also be used to estimate purity/impurity in a lot or sample when results are available. All of the capabilities of previous versions of Seedcalc (i.e., Seedcalc3, Seedcalc5, and Seedcalc7) are still available in Seedcalc8. They are described in two articles published in Seed Science Research from the June 2001 issue of Seed Science Research entitled "Statistical considerations in seed purity testing for transgenic traits" (June 2001 issue) and "Testing for adventitious presence of transgenic material in conventional seed or grain lots using quantitative laboratory methods: a new statistical approach and its implementation" (September 2005 issue). Seedcalc8 is enhanced with the ability to design qualitative testing plans using a Bayesian approach. This new capability has been presented at the 28th ISTA Congress 2007 under the title "A Bayesian approach for adventitious presence (AP) semi-quantitative testing in conventional seed lots".
This spreadsheet application contains macros that are necessary for the full implementation of the statistical tools. If the security setting of the computer is set at the "high" level, the macros will automatically be disabled. We suggest that the macros be enabled so that the user can realize the full benefit of this application.
Note that three earlier versions of Seedcalc (i.e., Seedcalc3, Seedcalc5 and Seedcalc7) are also available for download. While these earlier versions do not have as many features as Seedcalc8, they may run better under older MS Windows versions such as Windows 95.
By using both Seedcalc and Qualstat applications, testing plans can be designed for the majority of seed testing situations.
Please report any application issues or suggestions for improvements to Kirk Remund, Jean-Louis Laffont, or Sylvain Gregoire via e-mail.
SeedcalcStack9 - Jean-Louis Laffont (FR), Kirk Remund (US), Kevin Wright (US)
SeedcalcStack9 contains an implementation of the maximum-likelihood method for the estimation of the proportion of GM seeds with up to three stacked events in a conventional seed lot. SeedcalcStack9 was created in Microsoft® Excel, R, and the package RExcel. See the document "SeedcalcStack9 installation.doc" for guidelines on how to install all the components of SeedcalcStack9.
Please report any application issues or suggestions for improvements to Jean-Louis Laffont, Kirk Remundor, or Kevin Wright via email.
Qualstat - J.J. Daudin, C. Duby, P. Trecourt, J. Gregoire
Qualstat describes the efficiency of control plans when you check objects, and you count the number which is considered as defects to compare to quality specifications to accept or reject lots.
Qualstat also finds for you the optimal plan(s) if you give your quality criteria to meet, a good quality level that you wish to accept often (Acceptable Quality Limit or AQL) and a poor quality level you wish to reject often (Low-Quality Level)
It allows exploring the effect of false positive and/or false negative rates, as the check might not be 100% correct, due to the difficulty of the control, errors...
With Qualstat you can work with plans in which you check the samples individual by individual, or groups/pools. If you use pools, each pool has a positive/default or negative/ non-default result. The fact that a pool may contain more than one individual with the default is taken into account. Nevertheless, you have to make sure that your check allows you to detect at least one default individual in the pool.
Qualtstat can be used free of charge. Please refer to Qualstat if you use the results provided by the software in oral presentations, paper or electronic format. Authors disclaim any responsibilities for the use of the results provided.
Check for homogeneity when Unifying lots - Banyai J (HU), Dregelyi-Kiss A (HU)
When unifying seed lots to obtain a big compound lot, the main danger is heterogeneity.
The program performs a compatibility test which is an inverse heterogeneity test. Only those lots may be unified, from which the greatest difference of each quality characteristic value - purity, germination, other seed content - does not exceed a given limit. Such lots could be considered compatible.
The program indicates which lots are compatible or incompatible, and forms the possible homogeneous groups.
Box plots - Jean-Louis Laffont (FR)
The Zip file contains two Microsoft® Excel spreadsheets having a macro for creating box plots. The file having the .xls extension is for Excel 2003 users, whereas the file having the .xlsm extension is for Excel 2007 users. How to run these macros as well as a quick refresher on box plots is provided in these spreadsheets.
Please report any application issues or suggestions for improvements to Jean-Louis Laffont via email.
The Heterogeneity Testing Calculator for seed lots in multiple containers - Mustapha EL YAKHLIFI and Jean-Louis Laffont
A new update of the calculator, version 1.6, has just been launched and is now available for download. It fixes a bug in the calculation of the W statistic. Apologies for any inconvenience caused.
The Heterogeneity Testing Calculator is a Microsoft® Excel workbook which can be used to detect heterogeneity which would make a seed lot technically unacceptable for sampling for the issuance of an Orange International Seed Lot Certificate (OIC). The calculations carried out in this workbook are in accordance with the procedures published in the ISTA International Rules for Seed Testing, Chapter 2.9 (Heterogeneity testing for seed lots in multiple containers).