Monday: meetings, notebook

Coop & Moore 270: Species Trees

Discussing Bucky paper (Ane et. al. 2006)

Approach of BEST and *BEAST (last time): Prior probability on gene trees P(G|S), given species trees.  Goal: Posterior of species trees,

\[ P(S|X) = \int_G \Pi_{\ell} \left[ P(X_{\ell}|G_{\ell}) P(G_{\ell}|S) \right] P(S) dG \]

\[ P(G_{\ell}|S) = \Pi_{b \in S} P(H_b(G_{\ell})|S_b) \]

Uses a Dirchlet process model for clustering genes by topologies.  All genes in a cluster agree to share a common topology.  Comparisons to BEST and *BEAST next week.

Algorithms Group

Brief meeting today, discuss plans for next steps:

  • Start an OpenWetWare group to facilitate collaboration and sharing  of algorithms.

  • Dissect a common MCMC algorithm’s source code.

  • Identify some basic questions:  in MCMCMC how do we make account for proposal ratio?

  • Where do the priors enter: should we be using prior times likelihood in the Hastings ratio?

  • Next week: EM algorithms

  • Then: HMMs

Databases

Notes from Sunday

TreeBASE updates and to-do.

Notebook optimization

Checked profiling speed with https://gtmetrix.com/. Added wp-minify and started specifying sizes explicitly for images. Images from media library can be optimized for faster rendering using smushit, though most of my images are embedded from flickr instead, not sure how performance compares. Have been using wp super cache, but updated to some of the recommended settings.

Seems like further performance increases could best be accomplished with Content Delivery Network (CDN) option. Several plugins provide this. Use-google-libraries uses a couple public CDNs that google has made available for popular AJAX scripts such as jQuery. Super Cache works with the extension CDN sync tool, w3-total-cache provides CDN option as well.

Sounds like CDN for mathjax could be particularly valuable. Some description of this from the mathjax folks, not clear if they are providing the actual CDN or just the code to set one up? Not clear how to implement this on wordpress, but seeing if the knowledge-blog folks have any ideas.

References

  • Ane C, Larget B, Baum D, Smith S and Rokas A (2006). “Bayesian Estimation of Concordance Among Gene Trees.” Molecular Biology And Evolution, 24. ISSN 0737-4038, https://dx.doi.org/10.1093/molbev/msl170.