ValueError: sequence must be at least 20000 characters (17046 found)

[abrozzi]

Dears

I run:

chewBBACA.py AlleleCall -i ./data/raw/fasta -g ./gene_schema -o ./data/processed/gene --cpu 14

on 6285 assembly genomes of Klebsiella.

And I got this:

 CDS prediction
================
Predicting CDSs for 6285 inputs...
 [=====               ] 28% 27%
Error on predict_genome_genes:
Traceback (most recent call last):
  File "/home/user/anaconda3/envs/chewie/lib/python3.11/site-packages/CHEWBBACA/utils/multiprocessing_operations.py", line 42, in function_helper
    results = input_args[-1](*input_args[0:-1])
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/user/anaconda3/envs/chewie/lib/python3.11/site-packages/CHEWBBACA/utils/gene_prediction.py", line 217, in predict_genome_genes
    current_gene_finder = train_gene_finder(current_gene_finder,
                          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/user/anaconda3/envs/chewie/lib/python3.11/site-packages/CHEWBBACA/utils/gene_prediction.py", line 75, in train_gene_finder
    gene_finder.train(*sequences, translation_table=translation_table)
  File "pyrodigal/lib.pyx", line 5528, in pyrodigal.lib.GeneFinder.train
ValueError: sequence must be at least 20000 characters (17046 found)

Shall I pre-process the fasta files and filter out contigs <= 20000 characters ?

Bests,
Alex

[rfm-targa]

Greetings @abrozzi,

Thank you for your interest in chewBBACA. Regarding the error you have shared, are you using a training file to perform allele calling? It appears that chewBBACA is training a gene finder to predict genes from the input genomes, which should only happen if the schema does not include a training file or if you do not specify one with the --ptf parameter. We recommend using a training file to ensure the consistency of the gene prediction. The ValueError concerning the sequence length being less than 20000 characters may appear not because your genomes contain contigs with less than 20000 characters, but because one of the genomes has a total size less than that value. I recommend you check some assembly statistics for your genomes to make sure everything is within the expected values (e.g. genome size, contig number, GC content, etc).

Best regards,

Rafael

[abrozzi]

I confirm that using --ptf Kpn.trn it worked without that error message.

Thank you!

[abrozzi]

btw downloading on PC and uploading on Unix machine the https://github.com/B-UMMI/chewBBACA/blob/master/CHEWBBACA/prodigal_training_files/Klebsiella_pneumoniae.trn turns to be corrupted or incomplete.

It worked OK running directly prodigal -i reference_genome.fasta -t Klebsiella_pneumoniae.trn

HTH

[rfm-targa]

Hello @abrozzi,

Downloading files on Windows and transferring to Unix may lead to unexpected issues. I suggest you only do it if you are sure that downloading on Windows will not affect file format in any way or if you know how to convert them back to a format compatible with Unix.

Best regards,

Rafael