need exactly one. Specify --ksize or --dna/--protein.

[mw55309]

Hi

I am not sure exactly what this means.

I created a sig from reads using --scaled 10000 and -k 31

I then downloaded the k31 genbank database.

Now when I run:

 sourmash sbt_search genbank-k31.sbt.json AF-H4_S10_L001.sig -k 31

I get:

# running sourmash subcommand: sbt_search
When loading query from "AF-H4_S10_L001.sig"
2 signatures matching ksize and molecule type;
need exactly one. Specify --ksize or --dna/--protein.

What's happening? Same thing with sbt_gather

Perhaps as importantly, what command should I be running to compare reads from a metagenome against GenBank?

Cheers
Mick

[ctb]

Hi Mick, this looks like an old version of sourmash, so I'm not sure what bugs might be lurking. It looks correct to me, though. In the new version: sourmash search AF-H4_S10_L001.sig genbank-k31.sbt.json should work. It's possible that you are using an old version of sourmash with a new version of the SBT, for example. If you send me the .sig I can take a look at it to see if anything looks weird. best, --titus

[ctb]

(if you just want to get it to work without doing anything else, the only suggestion I have is to try adding '--dna' to the command)

[mw55309]

Hey Titus

Yeah we tried --dna and it didn't work

For the record, which installation instructions should we follow to get the latest stable version?

Cheers
Mick

[ctb]

This tutorial: https://sourmash.readthedocs.io/en/latest/tutorials.html will install the latest master branch, which should be as stable as it gets until we do a 2.0 release. best, --titus

[ctb]

Closing: I think the issue here was a version issue and should be resolved by installing the latest version. Will be addressed by doing "unstable" releases (see #324), also.

[nick-youngblut]

I seem to be getting this same error with with sourmash v2.0.0a2:

Command:

sourmash gather -k 31 -o tests/output_amy/sourmash/bins_DASTool_sourmash.csv tests/output_amy/sourmash/bins_DASTool.sig  /ebio/abt3_projects/databases/sourmash/genbank-k31.sbt.json > tests/output_amy/sourmash/bins_DASTool_sourmash.txt 	

Output:

...sig loading 5
When loading query from "tests/output_amy/sourmash/bins_DASTool.sig"
6 signatures matching ksize and molecule type;
need exactly one. Specify --ksize or --dna/--protein.

Adding --dna doesn't help. If I run sourmash compute on individual metagenome bins, then everything seems to work, but with the 6 metagenome bins used for input with sourmash compute in the example above, I get this error.

Any ideas on what's going wrong?

[ctb]

hi @nick-youngblut how was the bins_DASTool.sig computed? there's no simple way for sourmash to figure out which of the 6 signatures that are in there is the one to run gather on; for the moment the fix would be to put the signatures in individual files. Kind of a gap in the sourmash tool chain - eventually we'd like to be able to provide a signature name or md5 sum on the command line, but it's not there yet.

[nick-youngblut]

Thanks for the quick response! I computed bins_DASTool.sig with the following:

sourmash compute --scaled 10000 -k 31 -o tests/output_amy/sourmash/bins_DASTool.sig tests/output_amy/bin_DASTool_bins/*.fa 

The input was 6 fasta-formated metagenome bins.

[ctb]

On Sat, Feb 10, 2018 at 02:52:03PM +0000, Nick Youngblut wrote: Thanks for the quick response! I computed bins_DASTool.sig with the following: ``` sourmash compute --scaled 10000 -k 31 -o tests/output_amy/sourmash/bins_DASTool.sig tests/output_amy/bin_DASTool_bins/*.fa ``` The input was 6 fasta-formated metagenome bins.

OK! For now, I'd suggest just omitting the -o and it will produce 6 .fa.sig files in the current directory; then run gather on each of those. If you want to combine all six into one, you can either do cat *.fa | sourmash compute - -o output.sig -k 31 --scaled 10000 or (I think) sourmash compute *.fa --merged=someNameHere -o output.sig -k 31 --scaled 10000

…

-- I can check on that if you're interested but I'm on a loaner laptop right now :)

[nick-youngblut]

Great! Yeah, I'm trying to avoid multiple input, multiple output commands, which complicates snakemake pipelines.

I'll give the bin-file merging options a try. Will I still be able to differentiate the bins after merging, or will I just get a full list of taxonomic classifications for the entire merged set of bins?

[ctb]

On Sat, Feb 10, 2018 at 02:59:22PM +0000, Nick Youngblut wrote: Great! Yeah, I'm trying to avoid multiple input, multiple output commands, which complicates `snakemake` pipelines.

Ahh, I see. Apologies ;)

I'll give the bin-file merging options a try. Will I still be able to differentiate the bins after merging, or will I just get a full list of taxonomic classifications for the entire merged set of bins?

The latter, I'm afraid!

[sapuizait]

hi - i m having the exact same problem when trying to classify multiple contigs all in a single file
commands;
$ sourmash sketch dna -p scaled=1000,k=31 ~/tmp/shitflies/W22_Bacteria.filtered.fasta --singleton
$ sourmash search W22_Bacteria.filtered.fasta.sig gtdb-rs207.genomic.k31.lca.json.gz
$ sourmash search W22_Bacteria.filtered.fasta.sig gtdb-rs207.genomic.k31.lca.json.gz --dna

Output:
== This is sourmash version 4.7.0. ==
== Please cite Brown and Irber (2016), doi:10.21105/joss.00027. ==

select query k=31 automatically.
When loading query from 'W22_Bacteria.filtered.fasta.sig'
2311 signatures matching ksize and molecule type;
need exactly one. Specify --ksize or --dna, --rna, or --protein.

Is the solution to split in individual files and use gather? like mentioned above?
Thanks

[ctb]

yep! you can now use sourmash sig split to split many sketches into a directory full of files, or sourmash sig cat <input> -o somedir.d/ as well.