Megan6.6 versus 5.4 for taxonomic assignation

Hi !
In order to define the best LCA parameters for a taxonomic assignation I first used a control set of sequences (102 COI with corresponding known species).
These sequences have been BLASTn then a XML file generated under NCBI have been imported in MEGAN 5.4.0 or MEGAN 6.6.
I used the “nucl_gi2tax-Aug2016.bin” as taxonomic mapping file.

Using MEGAN 5.4.0 with the parameters (fig1), a perfect assignation was retrieved (in occurrence and frequency) : great !!
Using MEGAN 6.6 with what I suppose to be the same parameters (fig 2) the assignation was not accurate in particular for one clade for witch the assigned specie is not the correct one.
When I inspect the alignement of this clade I could see that the correct species have been filtered out despited it get the highest score. I suppose this because the special caracteres “?;” beside the score value (fig3).

I change some (all) parameters ( many tests performed) without success.
I wonder how to applied the same parameters in 6.6 as done in 5.4 in order to retreveid the same good results with Megan 5.4.0 I probably missed something ( does weighted % LCA still active), or if there is a bug in the latest version.
Could please help me this issue ?
Thanks
Best
Cédric



Dear Cedric,

looks like there is a problem determining the species for the given reference sequence.
Could you please send me the alignments obtained by uncollapsing the nodes that you show in figure 3 and then copying and pasting the alignments as text.

(This problem has nothing to do with the LCA algorithm or any parameters, the problem is associated with the mapping file used to identify the species associated with the reference sequence.)

Thanks
daniel

Hi Daniel,

Thanks for reply.
Please find the alignment in attachment.
Do you think that if the problem is associated with the mapping file, its mean that Megan 5.4.0 and 6.6 deal differently with this file…

Many thanks for your help

Best
Cédric

Inpector-node-brachy-rousseauxii (67.7 KB)

This problem was solved using the latest version of Megan (6.6.4 _21-12-2016) and custom taxonomic tre and map.

Thanks

Best regards
Cédric