Software Projects for 2006

This is the list of projects sent to students as possibilities for 2006. It is just for interest for the people who wanted to have some idea about the possible directions.

All the projects would suit two to three people, but we are flexible. All the projects will involve working very closely with a group member. Fortunately, they are all very friendly people.

Bayesian Classification - applications and extensions

Bayesian classification may sound obscure, but its proponents are more fanatical than shoppers looking for Schnäppchen after Weihnachts. We use the methodology for protein structure prediction and some unusual protein structure + sequence alignments. The project and extensions we have are

• parallelising a classification program with posix threads
• replacing one of the underlying distributions with more appropriate models
• removing some of the invalid statistical assumptions we make (independent variables)

Sequence design - dynamics in sequence space

Protein or RNA design are more interesting and important than Gesundheitssystem reforms. The existing literature is based on treating this as a discrete optimisation problem. We have a recipe for solving this by quasi-Newtonian dynamics in a fictitious sequence space. The aim of this project will be to implement a proof of concept version using a classic, simple protein model in a 2 dimensional discrete space. The coding will involve

• representation of sequence strings and the space
• implementation of a very simple energy function
• a simple integrator for the dynamics

RNA secondary structure drawing

There are many tools for drawing RNA secondary structures, but these are usually tightly coupled to prediction tools or very limited in their output. For both teaching and research, we have a serious need for a flexible tool which can provide various views (backbone H-bonds, circle plots, ..) and different output formats (bitmap, vector, ..) The work will have several aspects

• input parsing
• design of the internal data structures
• interface to various libraries for output

RNA sequence extraction from structure alignments

RNA sequence design is more fashionable than a Prada handbag, but perhaps only as effective as an Italian accountant. We would like software which can search for the most likely sequences for a structure, taking advantage of experimentally determined data. This can be done in either 2 or 3 dimensions and there several steps.

• initial alignments
• 2D - suffix tree representation and string search
• 3D - parsing of coordinates into a classifiable form
• clustering into sequence families
• automated checking using popular prediction codes

Mutant Wars, stochastic models and spectating

We have software which combines population modelling with evolutionary simulation, but are interested in extensions to the models as well as being able to interpret the results. There are two rather separate parts to this project

• Populations grow according to ordinary differential equations, but this does not fit well with Tsunamis, earthquakes, meteors and other events which kill small populations. The aim of this part of the projects would be to introduce formally correct stochastic differential equations
• When there are more than two participants in the mutant battles, it becomes rather difficult to see the results. The aim here is to build a smooth interface to a plotting library and be able to follow the life, death and evolution of the system.

Classical Dynamics Analysis

From molecular dynamics simulations, we have an urgent need for extracting some trajectory properties. Particularly in the interaction of DNA and proteins, we have to apply some measures to look at the kinds of interactions which take place and how important they are.