To create large scale protein alignment benchmark database.

Data is organized into 5000 and 10000 taxa simulations.

The following workflow was followed:

• Generation of guide trees:
  • r8s was used to generate topology (ultra metric tree).
  • Branch lengths were equalized and then adjusted randomly ranging from 1/4 to 4 and normalized to maximum tree depth of 1.0.
  • Appropriate parameters were added to create guide tree files for iSGv2.
  
  
• iSGv2 simulation:
  • 30 replications performed for each parameter set.
  • Parameters include:
    • Indel rate (0.0001 - 0.05)
    • Branch length scale (1x - 5x) [corresponding to substitution rates]
    • Gamma distribution (none or alpha = 0.1 - 0.5)
  
  
• iSGv2 simulation outputs:
  • MSA file
  • Sequence file
  • Tree used for simulation (including the scaled branch lengths)
  
  
• Analysis of results:
  • Maximum and minimum sequence lengths as well as the average alignment length were calculated for each of the 30 replicates.
  • Average percent gaps was also calculated for each of the 30 replicates.
  • Maximum likelihood phylogenies were reconstructed using FastTree 2.1.4 based on true MSAs, and missing branch rates were calculated between guide tree and FastTree tree using spruce 1.1 .
  • MAFFT v.6.864 MSAs were also generated for the first 6 replicates, and missing branch rates were calculated between guide tree and FastTree tree using these MAFFT MSAs.
  • The average, maximum, and minimum p-distances (the uncorrected difference divided by the number of alignment positions excluding gap sites) were calculated from all pairwise comparisons.