How to assign "Not assigned" BLAST hits by MEGAN

[javauma]

Hello experts,

I want to classify several millions of pathogen DNA/cDNA reads to taxonomical groups (like bacterial, viral, fungal ...).
My pipeline is as below...

Sequencer: HiSeq 2000 pair-end with NexteraXT library prep kit
Removing human reads: BOWTIE2
Mapping to nt/nr database: BLAST+ (-max_target_seq 1)

Then, I tried to classify BLAST output (tabular) to taxonomic groups by MEGAN5 with auxiliary mapping file they provided (http://www-ab.informatik.uni-tuebing...March2015X.zip).

This beautiful software successfully classified 90% of reads to specific taxons, but about 10% reads was "Not assigned".
However, I found some correctly BLASTed reads in this "Not assiened" group.
These reads are correctly assigned to specific pathogenic GIs and Taxon IDs, but "Not assigned" by MEGAN.
I guess this is because ..

(1) NCBI does not provide taxon ids to nt/nr database, so MEGAN miss certain amount of correct hits.
(2) MEGAN does not support latest gi-taxon database, and miss recently added GIs.

Thanks,

[javauma]

Finally, I updated BLAST+ to 2.3.0 which provides tabular outfmt with taxids.
And now I can classify all BLAST hits with taxids by MEGAN.
Thanks.

[SylvainL]

Hi, just a comment. I am not sure you can say your reads are specific here. If I remember well, setting "-max_target_seq" to 1 does not mean you will have only one match but that you will report only one...

[javauma]

Hi, SylvainL

I'm novice on bioinformatics, so I appreciate your kind comments !
To my understanding, you should hold as many hits as possible to improve plausibility.
(... so far as my computer permits)