- PASTA (Practical Alignment using SATé and TrAnsitivity)
is an improvement to SATé: it uses some of the
design of SATé but is faster, produces more accurate alignments
and trees, and can scale to much larger datasets.
PASTA computes alignments on very large
datasets using a divide-and-conquer
technique, as follows. It divides the dataset into smaller and evolutionary less diverged subsets,
gets alignments on those subsets, merges some pairs of these subset alignments to get a
set of overlapping and compatible alignments, and finally uses transitivity to
merge all these overlapping alignments and produce a final alignment. The novel transitivity-based merge
technique allows PASTA to be very scalable, but also improves its accuracy compared to
SATé, its predecessor technique.
Code: Available from github.
Image for MAC application is available here (version 1.5.0).
Refer to the README file on the github repository.
On MAC, you can simply download the image file, run it, copy the PASTA application to a location of your preference, and simply run the application.
On Linux, to run PASTA using GUI run:
Basic command-line usage is:
python run_pasta.py -i input.fasta -t starting.tree --auto
S. Mirarab, N. Nguyen, and T. Warnow, 2014. "PASTA: ultra-large multiple sequence alignment". Proceedings of RECOMB 2014
Supplementary materials are available
1000-taxon simulated datasets are available at SATé paper website.
Biological datasets can also be found at SATé paper website. However, we also used
thresholds other than 75% for biological datasets. The reference biological datasets without edge contraction can be found here.
RNASim dataset is obtained by creating random subsets of the RANSim dataset created by S. Guo, L.-S. Wang, and J. Kim and described here.
True alignments and tree are given here for our random subsamples.
All questions and inquires should be addressed to our user email group: email@example.com
| Copyright © 2013-2014 Computational
Phylogenetics Lab | ACES
3.304 | University of Texas |
Austin, TX 78712
Site help/questions/feedback/requests: e-mail Tandy Warnow