Full-length projects

David Ng, Identifying cyclic-nucleotide-binding proteins in Vibrio cholera

Cyclic di-guanosine-monophosphate (c-di-GMP) is involved in biofilm formation in medically interesting microorganisms. My project is to determine proteins in Vibrio cholerae that can bind to c-di-GMP; these proteins may be involved in biofilm formation. I will use bioinformatics techniques to identify proteins that possibly bind to c-di-GMP. I will then use wet lab techniques to clone the corresponding gene, express the protein, and test for the predicted binding activity.

Craig Lowe, Designing out the Cysteines of a Cysteine Knot

The C-terminal domain of the agouti-related protein is 45 residues and contains five disulfide bonds. This domain forms a structural motif known as the cysteine knot and is responsible for antagonizing many melanocortin receptors. This domain has been shortened to 34 residues and four disulfide bonds, while keeping the same structure and specificity for receptors. The domain is now being simplified further by removing all cysteine residues, with the intent of keeping the same functional signature and three-dimensional structure. The protein design is being done computationally and then will be verified experimentally.

Jes Frellsen, Using backpropagation in local-structure neural networks to do protein design

We investigate the use of backpropagation in neural nets to predict a primary structure of proteins given a desired secondary structure. The software package SAM-T02 contains a number of trained feedforward neural nets for secondary structure prediction of proteins. The input to these neural nets is a profile of the primary structure and the output is a probability distribution over a secondary structure alphabets, for example DSSP or STRIDE. We supply these neural nets with an input/output pair, where the output is the desired secondary structure and the input is a starting point for the primary structure. A primary structure, given the input/output pair, is then found using a backpropagation algorithm: the input is adjusted using the steepest descent gradient approach to minimize the error, which is the log-odds probability of the desired output.

Nathan Bahr, Evaluating multiple alignment methods

Coming up with a protocol for testing protein multiple alignments that does not suffer from rewarding overalignment.

Grant Thiltgen, Exploring New Alphabets

Implementing and testing new local-structure alphabets based on N-O hydrogen bonds.

Ron Chao, Genome-wide Prediction

Adding indexing to the whole-genome yeast protein structure predictions.

Tim Dreszer, Contact Lens: Towards a Better Structure Evaluation Tool

Rapid evaluation of the similarity of two conformations of a protein is an essential tool in protein structure prediction. There are a number of yardsticks currently used, each with its own advantages and disadvantages. The most widely used tool is RMSD, which suffers from several major shortcomings. In this project, I intend to investigate using a "contact" measure for structure evaluation. While not perfect, I believe it will prove more effective than RMSD as a rapid analytical tool for evaluating both global and local structure.

Lam Vu, Evaluation of secondary structure prediction

I will make a program (in Perl or c++) to evaluate the accuracy of secondary structure predictions. The input of the program will be the predicted values from different methods. The program will output the SOV, Qindex, and information gain measures.

Mini projects (for PhD students doing lab rotations)

Courtney Onodera, Identifying cyclic-nucleotide-binding proteins in V. cholerae

For my mini-project for this course, I will attempt to identify cyclic nucleotide (cNMP) binding-proteins from the bacterium Vibrio cholerae. cNMPs are involved in cell signaling pathways as "second messengers" that act intracellularly to mediate hormonal messages received externally. In particular, cyclic-di-GMP (c-di-GMP) has been shown to be involved in regulation of cell surface structures in certain microorganisms, ultimately affecting biofilm formation.

Alex Williams, SAM-T05 Web server

I chose the SAM-T05 web server project as a "small project" to work on alongside my rotation project. The primary goal of this project is to create an updated version of the SAM-T02 web site and scripts, from which the most recent SAM software can be run by web users. The final product of this project will include an updated web site, updated scripts, and merged features from the SAM-T02 files, the CASP6 files, and the yeast files. The final SAM-T05 site will include the updated prediction software of CASP6, the updated SAM-T04 template library, and the automatic image generation capability of the yeast web pages.

Sol Katzman, Predict-2nd: Protein local structure prediction with guide sequences

Preparing journal article on improvements to predict-2nd neural nets for predicting local structure.