Skip to content

Latest commit

 

History

History
30 lines (23 loc) · 2.39 KB

extra.md

File metadata and controls

30 lines (23 loc) · 2.39 KB
Raw

Additional considerations

Memory usage

Memory usage depends on the database. For a rough estimate of memory requirements for a given db, check the size of the hash.k2d file, which has to fit completely in memory. The manual recommends at least 175GB for the standard database. In my experiene, usage maxes out at the following values:

  • Genbank high quality: 64g
  • Genbank all: 256g

You should request at least this much memory on a node you plan to do classification on.

Time usage

Kraken2 is very fast compared to kraken1 or other classification tools. After loading the database in memory the first time (about 5 minutes for genbank hq), subsequet runs proceed very quickly. Using 8 cores on SCG I had the following results with the genbank hq database: 10399748 sequences (2356.32 Mbp) processed in 85.562s (7292.8 Kseq/m, 1652.36 Mbp/m). That's 85 seconds for a dataset with about 10 million read pairs. Obviously more cores can make things faster for larger datasets.

Classification percentages

I've evaluated some of the databases below on sets different metagenomic datasets, and compared the unclassified percentages to our old Kraken1 custom database. See this spreadsheet for a summary of results. Overall, Kraken2 with the high quality database classifies less reads than Kraken1 but much quicker. The full genbank database has increased classification percentages and speed.

Taxonomy modifications

When modifying the NCBI taxonomy with the improve_taxonomy.py script (which creates the taxonomy_array.tsv file necessary for downstream processing steps), the following changes are made to make the taxonomy more human interpretable:

  • Anything below a species gets set as rank subspecies
  • Prune the taxonomy to a specified set of levels.
  • To fix cases of "unclassified XXX" or "environmental samples" being at weird places in the tree and not having a useful name, use the heuristic:
    • if the name is "unclassified X" or "environmental samples" and everything below it is a species, and parent is not a genus or species than mark that as a genus
  • A few custom changes
    • Set crass-like viruses to genus
    • Set unclassified bacterial viruses to family
    • Set all superkingdoms to kingdom
    • Set Fungi and metazoa to be direct descendants of root