Skip to main content
Fig. 2 | Algorithms for Molecular Biology

Fig. 2

From: EMMA: a new method for computing multiple sequence alignments given a constraint subset alignment

Fig. 2

Experiment 0: MAFFT-linsi--add scalability issues. Alignment error (left) and runtime in hours (right) of MAFFT--add and MAFFT-linsi--add for adding 100, 200, 1000, or 2000 sequences to a 1000-taxon backbone alignment computed using MAGUS [13] on the INDELible 5000M2-het dataset with 5000 sequences. Averages over ten replicates are shown. Error bars shown for alignment errors are standard errors and standard deviations for runtime. We exclude replicate 4 because MAFFT-linsi--add encountered out-of-memory issues (64 GB) when adding 100 or 200 query sequences. Additionally, MAFFT-linsi--add is not shown for 1000 or 2000 query sequences because it either encountered out-of-memory issues or failed to complete within 12 h. Alignment error is the average of SPFN (fraction of true pairwise homologies missing in the estimated alignment) and SPFP (fraction of pairwise homologies in the estimated alignment not found in the true alignment). Results for SPFN and SPFP separately show the same trends and can be found in Additional file 1: Fig. S1

Back to article page