Error while loading rma file


#1

We have created a daa file using diamond version 0.9.10 with the following command:

diamond blastx --threads 20 --db nr --query $file --taxonmap prot.accession2taxid.gz --taxonnodes nodes.dmp --out $output --outfmt 100

taxonnodes option is useless apparently but anyway it was in the command we have launch.
As outlined in the diamond documentation we have used --outfmt 100 because it is supported by MEGAN and allows a quick import of results.

From the daa generated by diamond we have done a conversion in rma format using daa2rma.

daa2rma -i input.daa -o output.rma -at2 prot.accession2taxid.gz

We are using MEGAN Community Edition (version 6.10.5, built 23 Nov 2017).
The conversion seems to go fine and we get the corresponding rma file.
Then when we try to visualise the rma file we get this error:

error: Execute failed: java.io.IOException: Unreasonable string length: 1380794678

And no result seems to be loaded into MEGAN.
Could it be related to the size of the file ? The daa file is over 1 Gb (we didn’t use the -k option in diamond to limit the number of hit reported)

We will be grateful for any help, thanks

Jacques


#2

How you proceeded all looks correct, this appears to be a bug in daa2rma.

It would be better to use the program daa-meganizer, which does not compute a new file but rather does taxonomic and functional analysis and indexing of all the reads in your daa file and then appends the result of this to end of the daa file.
This file can then be opened in MEGAN.

This is not a size-based problem, RMA files can be many GB in size.