There are several issues with MUSCLE in the Cluster Analysis tool. The first issue is the inconsistency between versions MUSCLE v3 and v5 (v5 PPP and v5 Super5). A simple short test showed wildly inconsistent alignments between the three cases. The second issue is performance; the tests here were run using 750 sequences on a slow PC but show the differences between the modes. The legacy EFI tools used MUSCLE v3 which gave decent performance (7:35) with one thread. v5 PPP took 52:16 with 8 threads. v5 Super5 (should be used for larger alignments) took 2:48 with 8 threads. The third issue is deployment; MUSCLE v3 is not on GitHub and is subject to the reliability of the hosting server (www.drive5.com); the binary would need to be included in the repository in vendor/bin or in another GitHub repository specifically for dependencies. MUSCLE v5 is available on GitHub, both source code and Linux binary making it easy to deploy.
Due to the discrepancy in results a switch has been provided to allow the user to choose between versions 3 and 5, and the default is 3. Further examination of results is needed to decide on the proper version.
There are several issues with MUSCLE in the Cluster Analysis tool. The first issue is the inconsistency between versions MUSCLE v3 and v5 (v5 PPP and v5 Super5). A simple short test showed wildly inconsistent alignments between the three cases. The second issue is performance; the tests here were run using 750 sequences on a slow PC but show the differences between the modes. The legacy EFI tools used MUSCLE v3 which gave decent performance (7:35) with one thread. v5 PPP took 52:16 with 8 threads. v5 Super5 (should be used for larger alignments) took 2:48 with 8 threads. The third issue is deployment; MUSCLE v3 is not on GitHub and is subject to the reliability of the hosting server (www.drive5.com); the binary would need to be included in the repository in vendor/bin or in another GitHub repository specifically for dependencies. MUSCLE v5 is available on GitHub, both source code and Linux binary making it easy to deploy.
Due to the discrepancy in results a switch has been provided to allow the user to choose between versions 3 and 5, and the default is 3. Further examination of results is needed to decide on the proper version.