So I have noticed some incorrect assignment in MEGAN6 output. Allot of the alignment in the megan file could not be assigned to a taxonomy. The UBLAST was done using a 2015 nr datbase. I checked the accession in the alignment and it seemed that accession belonging to certain taxa changed (replaced) in the last year due to the updates NCBI did to the nr database. So I am wondering if it would be possible to provide me with the old mapping file ( dated before 2016).
I have already used the prot_gi2tax-Aug2016X.bin.zip before posting the bug report, but I had the same problem. I added a screenshot of the Megan alignment inspector to this post using the prot_gi2tax-Aug2016X.bin.zip mapping file. As you will see, there are several gray alignments (in which Megan could not assign a taxonomy), and if you took one of the accession numbers from these grey alignment (For example YP_007713606) and looked for it in NCBI website you will find that this accession was removed ( as seen here https://www.ncbi.nlm.nih.gov/protein/YP_007713606) . These grey unassigned alignments had much better e-values, and so a more accurate taxonomy assignments. I downloaded the March 2015 gi2taxa file from MEGAN5 webpage and used it, and this fixed the problem for me, however, on your webpage and also in other posts here (in Megan community ) you mentioned that this file has many bugs.