Data and results for: MAGUS: Multiple Sequence Alignment using Graph Clustering

This repository contains the datasets and corresponding results for the paper "MAGUS: Multiple Sequence Alignment using Graph Clustering". The Datasets.zip archive contains the ROSE, balibase, Gutell, and RNASim datasets used in our experiments. The Results.zip archive contains the outputs of running our methods against these datasets. Datasets used: ROSE: 10 simulated nucleotide model conditions from the SATe paper, each with 20 replicates, and with 1000 sequences per replicate. The ROSE datasets were originally taken from https://sites.google.com/eng.ucsd.edu/datasets/alignment/sate-i RNASim: This is a collection of simulated nucleotide datasets that were generated under a model of evolution that reflects selection due to RNA structural constraints. We sampled 20 subsets of 1000 sequences each, as well as 10 subsets of 10000 each, by randomly sampling from the original million-sequence RNASim dataset. Gutell: 16S.M, 16S.3, 16S.T, 16S.B.ALL: Four biological nucleotide datasets from the Comparative Ribosomal Website (CRW) with cleaned reference alignments from SATe. Since PASTA is restricted to datasets without sequence length heterogeneity, these were modified to remove sequences that deviate by more than 20% from the median length. The scrubbed datasets range from 740 to 24,246 sequences. The pre-screened 16S datasets were taken from https://sites.google.com/eng.ucsd.edu/datasets/alignment/16s23s BAliBASE: We use eight BAliBASE amino acid datasets used in the PASTA paper. As above, we remove outlier sequences, which leaves us with sizes ranging from 195 to 732 sequences. The pre-screened Balibase datasets were taken from https://sites.google.com/eng.ucsd.edu/datasets/alignment/pastaupp