Prokka -contig name to long even after --centre command

[Carola Berger]

I want to use Prokka to annotate a Velvet alignment of a Pseudomonas aeruginosa genome. When I try to run it I get this error:

Contig ID must <= 20 chars long: NODE_1_length_41061_cov_17.678381
[11:20:10] Please rename your contigs or use --centre XXX to generate clean contig names.

So I inserted the --centre XXX command in the running command. But now I become this error:

Contig ID must <= 20 chars long: gnl|XXX|PROKKA_contig000001
[11:22:08] Please rename your contigs or use --centre XXX to generate clean contig names.

So what do i do now? Is there a way to get the name shorter?

I appreciate any help

[maubp]

I would try renaming all the contigs in the input FASTA file before calling Prokka,

e.g. NODE_1_length_41061_cov_17.678381 --> contig000001

[Carola Berger]

I also got an answer for the developer team.

So maybe this helps other people who run into the same problem in the future:

1) Try using the "--compliant" option (and do NOT use --centre)

2) Or try "--compliant --prefix XX"

3) Or try "--compliant --prefix XX --centre XX"

[feralBiologist]

I tried using "--compliant" and "--centre XX" alone and in combination and it didn't work.

[Brian Bushnell]

Try downloading BBMap and running this command:

rename.sh in=contigs.fa out=renamed.fa prefix=contig

[Markiyan]

We use [strain#]_AS[AS#]_CO[contig#]

It is extremely important to have both descriptive and consistent naming schema for the contigs/scaffolds for all downstream analysis.

Unfortunately NCBI names are a bit too long and usually have white spaces before the contig#, which makes them unsuitable as human and machine readable fasta_ID...

In our case we usually use the following contig names:

>[strain_name]_AS[AS#]_CO[contig#] {some optional description/orig name/etc}*

like:

>NRRL2338_AS1006_CO1

in case of scaffolds we put SC instead of CO,

>NRRL2338_AS1006_SC1

so if you has assembled DH10B and yours assembly #5 has multifasta has:
>contig0001
...
>contig0012

you do search and replace
>contig00
with
>DH10B_AS5_CO

PS: in the case of long strain names they may need a bit of shortening.

contigs renaming happens after assembly selection/polishing.

For the data downloaded from the NCBI/EMBL, it is done in KATE or similar text editor supporting regexp. Also one can use perl one liners (google - perl search and replace) if you are familiar with perl regexp.

Being consistent in sequence naming has a huge impact in all types of downstream analysis (blast, mapping, annotation).