CSCI 6904 Genomics and Biological Computing Seminar 1

CSCI 6904 Genomics and Biological Computing Seminar 1 – Sequencing a genome in one step. Projects and paper presentations

Genetic engineering, four slides about it. It is possible to cut and paste DNA into artificial constructs: • Mutate on demand • Make hybrid • Neutralize toxic genes by expressing them in two parts Michael Smith UBC, Nobel 1993 in chemistry

Ctrl-X Ctrl-V on DNA Principle Palindrome Sequence Cut and leave “Sticky ends” Isolate on the basis of size Paste somewhere useful Why this exist? To systematically destroy foreign DNA (Virus, parasites, etc…).

Genetic engineering Plasmid Artificial construct used to manipulate sequences. Cloning Make a copy of a segment of DNA

Genetic engineering Mutant Sequence in which an Copy error is introduced during DNA replication. By the way… Everybody is a mutant relative to their parents.

Genetic engineering Site directed mutagenesis Mutation can be introduced on demand by using artificial DNA fragments with one/a few error in them.

Genetic engineering Example 1. Environmental Bioremediation 1. PCB degrading bacteria 2. Oil eating bacteria 3. Plastic degrading bacteria maybe someday. 2. Genetically modified organisms 1. Human insulin producing pigs. 2. Antifreeze produce. 3. Long shelf-life tomatoes. 4. High-fat or specialized fat grains. The bulk of the ethical issues may be largely due to a natural resistance to new technologies.

Polymerase Chain Reaction Principle Exponential amplification of a single piece of DNA. • Forensic science -> DNA evidence. • Rapid medical diagnistic • Jurassic Park (dinosaure blood in amber samples)

DNA replication Needs (in vitro): -template DNA -Polymerase -Primers -Nucleotides in solution

Polymerase Chain Reaction Principle Exponential amplification Taq

Paper presentation Problem Why this is a problem of interest. Abstraction Translate your problem into something that has nothing to do with biology anymore. Setting the “specifications” or the problem. How it was done Methods, results and conclusions. Validation

Geno. Frag Problem Need to amplify rapidly a whole genome to quickly identify the difference. This can be particularly useful in case of outbreak which cause is still unknown coming from a bacteria/virus not know to have such harmful characters. Example of this would be: E. coli contamination of watershed (Walkerton, June 2000). Flesh-eating disease (S. aureus) SARS HIV Influenza General case of drug resistance.

Geno. Frag Abstraction Need to be able to scan whole genomes very fast. Assume that the target organism is very similar to something we already know the genomic sequence. Target of interest: S. aureus. An ubiquitous, yet sometimes extremely harmful bacteria (mutation, inversion, deletion, transposons, …). Technique: Long range PCR. Problem: need a lot of oligonucleotide primer pairs to cover the entire sequence.

Geno. Frag Problem with primers 25 nt long Given G+C content No character repeat longer than N No “hairpins” Correct Tm. Self- and intercomplementarity Unique (Representative of all sub-species) Sufficiently informative Equally spaced Maximally covering all the dataset

Geno. Frag Problem with coverage pairs must be roughly equally spaced (9 -11 Kbp). Must overlap to some extent. Minimize the number of pairs. ALGORITHMS: Shortest path problem. Single Traverse Algorithm.

Geno. Frag Problem with primers 25 nt long Given G+C content No character repeat longer than N No “hairpins” Correct Tm. Self- and intercomplementarity Unique (Representative of all sub-species) Sufficiently informative Equally spaced Maximally covering all the dataset

Geno. Frag The filtered potential primers are reasonably well distributed along the DNA sequence.

Geno. Frag Optimizing SSP - 10 K fragments - Maximal coverage SITA - Equally sized fragments - Maximal coverage

Geno. Frag Graph abstraction of the problem

Geno. Frag Graph abstraction of the problem Hence the linearity of the problem. SSP graph

Geno. Frag Graph abstraction of the problem Hence the linearity of the problem. SITA graph

Geno. Frag Results Performances - 40 s for 2. 8 Mb on 1. 6 GHz PC Failed when… - Presence of large insertions** - Less general use if omit to use filter 7. But otherwise: - 1 -step genome (or part of one) amplification seems possible.

Paper presentation What to choose: Try to do a search on a method or field of your choice and add bioinformatics in he query. There will be a selection of papers that will be related to something you will enjoy talking about. Examples Bioinformatic resources (database, interface, services) Large scale projects (Folding@home, Blue. Gene, HGP, ) API for bioinformatics (Bioperl, Biopython, NCBI tk, etc…) Machine learning applications (detection, prediction, method validation) Parallelization computing in bioinformatics Algorithm and application to a specific question Theoretical papers and simulators * Methods and applications * *would require a bit more background knowledge.

Projects ideas A protein contact map using Voronoi triangulation Problem Proteins have fairly complex 3 D chain paths. Many structural bioinformatics methods require the knowledge of which characters are in contact with each other. Such contact maps are usually implemented as applying a cutoff filter to a distance matrix between arbitrary chosen centroids. A contact map based on whethere is a shared surface in a Voronoi diagram between two amino-acids would be a nice, general purpose bioinformatic tool. References http: //bioinformatics. oupjournals. org/cgi/content/abstract/bth 365 v 1

Projects ideas A lattice-based protein folding simulator Problem Protein are computationally expensive to model using computational chemistry methods. Further, these models are empirical and have a rather limited scope which does NOT include protein folding. There is a 2 D/3 D abstraction of protein chain that exist, fixing each amino acid to one vertex in a cubic lattice. With this simulation environment, seriously cool experiments can be performed, especially if the environment is efficient. This type of simulation is actually simple, compared to the simulation environment of phyiscal systems. References

Projects ideas Finding phylogenetically informative sites using Matrix decomposition (PCA, ICA, …) Problem An in-house phylogenetic application is generating solution pools of phylogenetic trees. This creates matrices of trees columns and sites (datapoints as rows) There is an interest to figure out which rows are the most informative at discriminating between trees, or whether it is possible to identify clusters of data points showing some level of dependence to each other (They are assumed to be independent). Identify such cluster could be used to identify regions of recombination, for example. References

Projects ideas Computing the shortest path of topological rearrangements between two binary tree topologies Problem Information about the optimization landscape in phylogeny is scarce. Drawing paths between solutions would allow to plot cross-section of the search space to access the shape of the search space in various locations. References Felsenstein, Inferring phylogenies, 2003, Sinauer Eds.

Projects ideas Sequence Harvesters Problem Gathering sequence information from Gene. Bank is a time consuming task. Typically, one starts with a sequence of interest, query the database with BLAST, chose the laergest set of non-redundant sequences and paste them into a file. The sequence have to be renamed one at the time and checked for duplicates. This could be automated and run in a few seconds (instead of a few hours of fingerwork). A suggested platform for this project would be Biopython. References See www. biopython. org

Projects ideas Using XML for validating protein structures Problem Protein 3 D structures are stored in files in the PDB format. This format is regularly abused. As a results, it is hard to parse directly the files. There are cases of missing information, duplicates of atoms, omission of labels, non-standard labels. This project would explore the use of the new PDBML format to use as data source instead of the PDB flat files. References See PDBML from www. pdb. org The main ref isn’t available yet to Dalhousie. . .

Projects ideas This is not a closed list! Text Mining and conceptual Biology. Gene expression and clustering. Protein topologies and similarity detection. Clustering 3 D structures. Sequence patterns near variable regions in proteins sequences.