Workshop on Probabilistic Modeling and Machine Learning in Structural and Systems Biology (PMSB), Tuusula 2006
The ever-ongoing growth in the amount of biological data, the development of genome-wide measurement technologies, and the shift from the study of individual genes to systems view all contribute to the need to develop computational techniques for learning models from data. At the same time, the increase in available computational resources has enabled new, more realistic modeling methods to be adopted.
In bioinformatics, most of the targets of interest deal with complex structured objects: sequences, 2D and 3D structures or interaction networks. In many cases these structures are naturally described by probabilistic graphical models, such as Hidden Markov Models, Conditional Random Fields or Bayesian Networks. Recently, approaches that combine Support Vector Machines and probabilistic models have been introduced (Fisher kernels, Max-margin Markov Networks, Structured SVM). These techniques benefit from efficient convex optimization approaches and thus are potentially well-scalable to large problems in bioinformatics.
The increasing amount of high-throughput experimental data begins to enable the use of these advanced modelling methods in bioinformatics and systems biology. At the same time new computational challenges emerge. Statistical methods are required to process the data so that underlying potentially complex statistical patterns can be discerned from spurious patterns created by random effects. At its simplest this problem calls for data normalization and statistical hypothesis testing, in the more general case, one is required to select a model (e.g. gene network) that best explains the data.
The aim of this workshop is to provide a broad look at the state of the art in the probabilistic modeling and machine learning methods involving biological structures and systems, and to bring together method developers and experimentalists working with the problems.
We encourage submissions bringing forward methods for discovering complex structures (e.g. interaction networks, molecule/cellular structures) and methods supporting genome-wide data analysis.
Find out more at the workshop website.