Could the MEGAN5 feature of LCA of a fixed percent be implemented in MEGAN6 in addition to the weighted LCA algorithm? That way depending on the research question you could choose one or the other. We find it useful to allow some potentially mis-annotated genbank sequences to hit a read and have it still be assigned a taxon since the organisms we target are not highly studied. I would use MEGAN5 instead but MEGAN6 also does a better job at interpreting taxon names in NT genbank, as well as interpreting results from the SILVA database.
I have (re)-implemented this feature and it will be available in the next release of MEGAN 6.
You will see that the option Percent to cover is now enabled when the LCA algorithm has been set to naive. If you set the Percent to cover value p to less than 100%, then each read will be assigned to the LCA of p-% of the taxa to which it has significant a alignment.
Will this update be available in a release of MEGAN6_v8 or will it have to wait until MEGAN6_v9? I just checked the latest release of v8 and I don’t believe it was updated yet.
I do keep some of the older versions on the web server (I can also put very old versions back there on request). Version 6.8.20 is still on the web server, so just replace 6_9_0 by 6_8_20 in the URL.
Also, please let me know what problem there is with the latest version that make you want to go back to an older version.
We want to use MEGAN6 on a local server which is used by a couple different groups and the administrator was hesitant to install a new version 6.9 when it had just been released and perhaps not tested by many different users. We will likely update to 6.9 after another couple months.