This directory contains files pertaining to the analysis of marginal topology probabilities by summing over alignments contained within a DAG, as described by Herman et al. (2015).

The alignment samples contained in `4_globins.log' and tree samples in `4_globins.tree' were generated by running StatAlign v3.2 on four globin sequences (human cytoglobin, myoglobin and α-haemoglobin, as well as lupin leghaemoglobin) for 200,000 iterations using the Dayhoff rate matrix, taking samples every 100 iterations. 

These 2000 alignment samples were then further thinned down by a factor of 20 to yield 100 alignments for the purposes of subsequent analysis.

The marginal probability of each of the three possible tree topologies was computed by averaging over 500 of the tree samples in `4_globins.tree', as discussed by Herman et al. (2015). These marginal probabilities were computed on each of these alignment samples individually, as well as on the DAG formed from the 100 alignment samples.

The script `example-analysis.sh' carries out the above analyses using the program WeaveAlign (available at http://statalign.github.io/WeaveAlign), requiring 1-2 minutes on a 2.3GHz core. 

References
^^^^^^^^^^
Herman JL, Novák A, Lyngsø R, Szabó A, Miklós I and Hein J (2015) "Efficient representation of uncertainty in multiple sequence alignments using directed acyclic graphs." BMC Bioinformatics (to appear)
