So far, three extensive studies (Armougom et al., 2006a; O'Sullivan et al., 2003; Pei et al., 2008) have suggested that the results obtained with these alignment-free benchmarking methods are in broad agreement with those reported when using regular benchmarks.
Two WebLogos were created with these alignments.
Construction of DMs started with these alignments.
Working with these alignments, we also obtained very similar results and conclusions regarding heterotachy, compositional heterogeneity, misfit analyses, and bootstrap support.
Running Cuffdiff and FDM with these alignments somewhat decreased FPs on the replicate pair tests, but did not markedly change recall and precision on the non-replicate pair tests with a target FDR of 0.05 (Supplementary Table S2 and Supplementary Fig. S6).
The amino acid sequences were aligned using T-COFFEE 7.81 [31], the ambiguous regions within these alignments filtered with GBLOCKS 0.91 b [32], and the filtered individual sequences concatenated.
To further compare individual aligner with these two alignment algorithms mentioned above, we performed evaluation and comparative analysis of these aligners in terms of computational performance, alignment accuracy, and application-specific features.
The researchers also repeated these alignment simulations with a woodpecker version of their model because woodpeckers are known for being able to withstand extreme head-on accelerations, which can be 10 times greater than those that cause severe concussions in athletes.
We then used the codeml package from PAML (v.4.9a) on each of these alignments with the WAG amino acid rate matrix to calculate maximum likelihood genetic distances between the two sequences40.
We then compared the alignment scores obtained from these alignments with those seen for our orthologous and paralogous gene pairs.
The evolutionary tree in Figure 1 was constructed using the HKY+gamma model from these alignments with gap positions removed.
