Notes 20101102 BIOL 614 Bioinformatics Tools
From SnOwy - Ed's Wiki Notebook
Contents |
Presentations Nov 16
- 20 minutes + 10 discussion
- emphasize biological background
- examples in the tools
- project submission Dec 10
- Me: Seminar 8; Tuesday Nov. 23
- Send paper to Brendan at least 3 days ahead
Sequence to Structure Software
- we discussed multiple sequence alignments onto structures
- swiss model -- structural modeling
- does a structural search instead of a sequence search
- starts with a MSA profile to start a search with corresponding structures
- Swiss-Model is a homology modelling server -- it's intermediate as it's looking for a similar structure
- -- this is instead of making a de novo structure from a given primary sequence
- other examples of servers ...
- fold recognition server instead -- instead of looking for homology, look for pieces of sequence with known structure
- secondary structure prediction
- fold recognition 3D-pssm -- distant homologues / multiphyletic
Example: ALS
- degenerative disease
- 3 to 6 year
- looked at the amyloid plaques and found two proteins had aggregated
- included TARDBP (TDP43), SOD1
- inclusion bodies
- inclusion bodies were phosphorylated and ubiquitinated
- found many glycines and serines in the C-terminus
- increased probability of phosporylation (motif)
- FG repeats create loose lattices with high selectivity as nuclear pore gels :D COOL!
- some cysteines?
- an ordered alpha-helix with cysteins near the end of the C-terminus
- cysteins stack with aromatics perhaps?
Functional Genomics -- Microarrays
- genomics: characterization of genome -- static aspects -- sequences
- functional genomics: describe interactions, dynamics, changes, functions
- microarray -- could this be replaced by high throughput next-gen sequencing technologies?
- can be used for proteomics
Microarrays
- removing system bias
- Cy5 (red) labaled probe fore healthy tissue as a control for expression profile in Cy3 (green)
- Inferential Stats: T-tests, False Discovery Rate (FDR)
- Descriptive Stats: Clustering
- goal -- biological variation
- assume that we can normalize things -- that within a single experiment, the relative gene expression levels stay the same
- should normalize Cy5 and Cy3 -- sometimes either the red or green channel will appear much brighter as an artifact
- often yields a plot with much low expression and very limited high expression of genes
- using a log scale transforms makes this more useful
