When starting with a laboratory sample, the software allows the considerable reduction of costs for the control of GMOs while keeping the agreed risks of false results sufficiently small. It also helps stakeholders to investigate the most appropriate detection method, including new techniques such as microarrays or whole genome amplification for detecting both approved and “unknown” GMOs. However, the programme has a wide scope beyond GMOs, if analyses on other food-related factors are required, such as pathogens, mycotoxins, or allergens.
Let me explain that with an example of a feed producer. This stakeholder receives a new maize lot, and would like to know whether he should accept or reject this new shipment. He would like to know if GMOs are below or above a given labelling threshold. An initial sample is taken in compliance with ISO or other standardised sampling plans, and is sent to the laboratory for analysis. The "acceptance sampling plan" then defines the procedure on how to “handle” this “laboratory sample” with the aim of keeping risks below predetermined, low levels while remaining cost-efficient. This is a worldwide accepted method, based on statistics and widely used - for example, by seeds companies worldwide - and which also can be applied to kernels. This sampling plan is of particular interest in cases of a potential heterogeneous distribution of traces in a lot, or when protein based methods, such as the SDI detection kits, are used.
OPACSA utilises the very robust way of designing an "acceptance sampling plan" by applying qualitative, generally more cost-effective tools in the lab rather than quantitative, expensive and complex measurements. The stakeholder provides relevant input and, based on this information, the programme then finds an adequate solution minimising the costs. The software also allows the obtaining of sampling plans for either one or two analysis steps, which are referred to as single or double sampling plans. As indicated by its name, double sampling is conducted in two consecutive steps. A few sub-samples first are analysed. A second set of sub-samples may be then assayed, but only when the first leaves some doubt. Since the first analysis is conclusive in many cases, double sampling can achieve the same control with a far lower cost. The optimisation process is more complicated, although this is just a matter of a slightly longer computer calculation time for the end-user.
Stakeholders have to provide the thresholds negotiated by both seller and buyer and the corresponging acceptable error risks. They also have to give the cost of the kernels, expressed relatively to the assay cost, and to provide the genereally expected GMO, or searched analyte, percentage. This percentage, generally for below the threshold, is used to get the expected cost to minimise. These elements are then sufficient to launch the computation that analyses the data and, in order to find the most suitable sampling plan for any given trading partner, it is possible of course to make successive computations with several slightly different hypotheses. The software was designed to minimise the time of the computations, which in most cases take only a few minutes.
The programme is unique because it is the first software that takes into account important economic issues. These factors are not negligible, especially in cases in which the detection methods, or materials to be analysed, are expensive. In contrast to other programmes such as QUALSTAT and SEEDCALC, developed by the “International Seed Testing Association”, OPACSA achieves the same error risk, but with half, or even less, of the costs for the analyses.
The programme is valuable for each food chain starting from “discrete attributes”, such as kernels. It is also of interest for laboratories working with expensive detection systems, such as micro-arrays, quantitative or multiplex PCR, or for stakeholders trading with expensive seeds, for example. It is quite likely that OPACSA initially will be used by seed companies, due to their worldwide use of such sampling control plans. However, enforcement authorities may also benefit by using OPACSA to make the best of their limited budget when facing increased demand for analyses.
We hope that the programme will lead to a higher acceptance of the qualitative and affordable methods in the GMO detection world, where, up to now, quantitative and expensive methods such as the real-time PCR are the standard. This runs in parallel with the development of new qualitative multi-targets systems such as qualitative multiplex PCR carried out in the work packages 5 and 6 of Co-Extra to overcome technical limits of quantitative PCR, or new microarrays by Eppendorf Assay Technology SA.
The software uses the PYTHON language, which is freely available on the net. The OPACSA software is freeware, made publicly available by INRA within the framework of the Co-Extra project. The source code is public as well, which allows free adaptation of the software to different needs of the stakeholders, under the prerequisite that its origin clearly be stated. There is no need to know PYTHON to use the software, but it can be useful to learn a bit of this language to modify the software for specific needs.
I am currently waiting on feedback from software users to further improve the programme. We also will use the forum facility of the Co-Extra site to answer questions that may arise, as well as to retrieve requests for software enhancements. The main request probably will be an improved user-friendly interface, which is outside of our commitments but can be developed by third parties. I am also in touch with a stakeholder who is in charge of the controls at a seed company. He has asked me for some small complements. Once added, these also will be made publicly available.