mSWEEP supports using a custom reference database. A typical workflow for constructing the custom database might proceed as follows
-
Gather assembled sequences for the species of interest. If you are unsure of what species are present in your sample. use taxonomic profiling tools like MetaPhlAn2 to identify them.
-
a) Provide a grouping for the assemblies (e.g. sequence types, clonal complexes, or the output of some clustering algorithm.). PopPUNK is the recommended choice but if your species have an established multi-locus sequence typing scheme, that can also be used.
-
(Optional) Filter out reference sequences that cannot be reliably assigned to a group (eg. the sequence type cannot be determined) and perform other appropriate quality control measures. We recommend filtering assemblies using checkm with a completeness threshold of >90% and contamination <5%.
-
(Optional) Set up the demix_check index for additional QC checking of the bins produced by mSWEEP/mGEMS.
-
Index the database with Themisto.
For prebuilt reference databases, please refer to the supplementary datasets in the article https://www.nature.com/articles/s41467-022-35178-5.