HERMeS: A second generation approach to the automatic analysis of two‐dimensional electrophoresis gels. Part III: Spot list matching

Abstract

Abstract One of the most critical steps in two‐dimensional gel analysis is the comparison of spot lists. In this paper, we describe our approach to the matching problem. We propose a method based on syntactic pattern recognition techniques, which differs from previously used methods and offers a major improvement to cope with discontinuous geometric distortions. The method is based on non‐metric considerations and describes the pattern observed on a given gel in a non‐geometric way. This approach leads to the definition of finite automatons which can be used to automatically find the presence of a given pattern on a gel. The program uses a technique of clause generation related to artificial intelligence techniques. A set of rules is applied to the definition of homologousspots, and probability coefficients are computed to compare and to chose between different pairing hypotheses. Once matching has been achieved between a set of gels, the construction of a master gel which summarizes the information obtained is allowed. Two types of masters can be used. The first one summarizes the information between spots lists belonging to the same experiment and includes intensity considerations, while the second one summarizes the information resulting from different gels and does not include spot intensity and volume. Masters are written and known by the system as a spot list. This ability allows them to be handled just like spot lists and, using them, to perform all the actions available to the system. This approach is compared with a more classical one based on geometric corrections. The use of probability coefficients is also discussed and extended to their application to multiple matching and further utilization of masters.