# Reconstruction Plan Spring09

From Biowiki

2 major programs - enhancing and testing handalign, analysis of reconstruction accuracy

### flowchart (why not!)

### enhancing and testing handalign

- parameter sampling (in progress)
- autocorrelation (determine best rates of sampling moves to maximize mixing/time ratio)
- rerun balibase benchmarks
- run benchmark analysis

- figure out a way to partition sample runs and create consensus alignments
- want to find a way to create representative alignments for each 'peak' in distribution
- partition sample runs by log likelihood
- group alignments within partitions using AMA
- create consensuses from resultant groups using CASA

- reconstruct something (DHPS, cathelicidans, ...?)

### analysis of reconstruction accuracy

- finish mathematical analysis
- create simulated cherries, etc. Use multiple reconstruction(/simulation?) methods. Compare to mathematical results.
- run Tree Fam data, compare with simulated and mathematical results.
- build 'representative' benchmark set

-- Lars Barquist - 28 Jan 2009