Frequently asked questions#

Here you will find answers to some common questions about geNomad. If you can’t find an answer to your problem here, please open an issue in the GitHub repository.

How can I speed up geNomad?#

If you want to speed up the execution of geNomad, there are two options available:

  • Disable the neural network-based classification using the --disable-nn-classification option, which will also disable score aggregation and force geNomad to solely rely on the marker-based classifier.

  • Decrease the sensitivity of the MMseqs2 search that assigns markers to genes with the --sensitivity parameter. This will make the annotate module faster, but will also decrease the number of genes assigned to markers.

Please note that both options may negatively impact geNomad’s classification performance.

How can I get the taxonomy of all my sequences, regardless of their classification?#

If you are only interested in performing taxonomic assignment of viral genomes and not classification of the sequences, you can run the annotate module alone using the following command:

genomad annotate input.fna genomad_output genomad_db

The taxonomic assignment of your sequences will be in the genomad_output/input_annotate/input_taxonomy.tsv file. Please note that the path may vary depending on the name of your input file, but the structure will remain consistent.

Why is the execution is stuck at the “Classifying sequences” step of the nn-classification module?#

During the neural network-based classification of sequences, the processor performs numerous computationally expensive matrix multiplications. Although modern processors can handle these computations quickly, some hardware, particularly older or laptop processors, can be quite slow during this step.

If you ran geNomad using the end-to-end command and your execution is stuck at the “Classifying sequences” step of the nn-classification module, you can disable the neural network classifier by using the --disable-nn-classification option. By doing this, geNomad will rely solely on marker-based classification and disable score aggregation.

Why execution is failing at the “Annotating proteins with MMseqs2…” step of the annotate module?#

If geNomad fails during the “Annotating proteins with MMseqs2 and geNomad database” step, it is likely that your computer is running out of memory. This can be resolved by using the--splits parameter, which splits the marker database into smaller chuncks and searches each of them independently.