nhmmer failed to run - Error: Invalid alphabet type in target for nhmmer. Expect DNA or RNA

[minjinhan]

Hello
I try to run barrnap to identify rRNA from a eukaryotic genome , the commad as follow:
barrnap --kingdom euk --threads 20 --outseq rRNA.fasta < chr1.fasta

After running, we got following error . Can you supply suggestions to solve this problem? Thanks!
[barrnap] This is barrnap 0.9
[barrnap] Written by Torsten Seemann
[barrnap] Obtained from https://github.com/tseemann/barrnap
[barrnap] Detected operating system: linux
[barrnap] Adding /miniconda3/lib/barrnap/bin/../binaries/linux to end of PATH
[barrnap] Checking for dependencies:
[barrnap] Found nhmmer - /miniconda3/bin/nhmmer
[barrnap] Found bedtools -/miniconda3/bin/bedtools
[barrnap] Will use 20 threads
[barrnap] Setting evalue cutoff to 1e-06
[barrnap] Will tag genes < 0.8 of expected length.
[barrnap] Will reject genes < 0.25 of expected length.
[barrnap] Using database: /miniconda3/lib/barrnap/bin/../db/euk.hmm
[barrnap] Scanning chr1.fasta for euk rRNA genes... please wait
[barrnap] Command: nhmmer --cpu 20 -E 1e-06 --w_length 3878 -o /dev/null --tblout /dev/stdout '/miniconda3/lib/barrnap/bin/../db/euk.hmm' 'chr1.fasta'
[barrnap] ERROR: nhmmer failed to run - Error: Invalid alphabet type in target for nhmmer. Expect DNA or RNA.

I am sure there are no other alphabets in the fasta sequence except A/T/C/G.

[snayfach]

I've gotten the same error. For me, what caused the error was one sequence composed entirely of G and T nucleotides. Adding a single A and C nucleotide resulted in no error. This should be an easy :-)

[jdwinkler-lanzatech]

I also just ran into this problem as well.

[zxgsy520]

I also just ran into this problem as well. I added A and same problem. The Internet said it was a problem with the conda installation.

[ptrebert]

I just stumbled upon this; in case this is still relevant @zxgsy520 there is a switch to set the alphabet type for the query and use this as "guide" in case the alphabet type cannot be guessed for the target; --dna introduced in this PR
EddyRivasLab/hmmer#252
The switch is available in nhmmer v3.3.2 installed via conda

correction: the fix in the PR has only been merged into the dev branch, the switch --dna exists in latest release but does not include the fix

[ZeweiSong]

I bypassed this issue by replacing all ambiguous bases (M, K, H, et al.) to N.

[cabbagesofdoom]

I got this issue for a genome that started with a telomere repeat and did not have all four bases in the first few hundred characters. I got around it by replacing the first four characters of each sequence with GATC and then running on the temporary file:

PREFIX=$(basename ${GENOME/.fasta/})
sed 's/^[ACGT][ACGT][ACGT][ACGT]/GATC/' $GENOME > $PREFIX.tmp.fasta

[HansongYan666]

hello everyone:
I met the same error, cab's @cabbagesofdoom method is work, just replace 4 base, I suggest to release the hmmer version, first, scan the hmm file in barrnap db directory, for example euk.hmm, the hmm file has a version info in the header, uninstall the hmmer and install the same version hmmer, then it works on the sample genome. tips: my genome only with bases ATCGN. you can replace other base first.

[acontrerasg]

The fix above worked for me as well.

To provide further clarification:

1. I verified that the HMMER version for euk.hmm in the directory /home/user/miniconda3/envs/barrnap/lib/barrnap/db/ was 3.1b2.

2. I manually downloaded the HMMER package from Bioconda: HMMER 3.1b2 files.

3. I installed the package in the conda environment where barrnap was located using the following command:

   mamba install '/home/user/downloads/hmmer-3.1b2-3.tar.bz2'
   

(Note: You can use conda instead.)

After completing the installation, I ran barrnap on my assembly, and the issue was resolved.

PD
My assembly contained telomeres in chromosome 1 with the sequence "CCCTAAA". This seems to have caused the error, the lack of a G nucleotide in the sequence for the first few hundreds bases.