There is a separate analysis application for detecting multiarch related bugs.
It has the same requirements as the dedup application. It uses four files:

    multiarchschema.sql
    multiarchimport.py
    multiarchanalyze.sql
    multiarchanalyze.py

With the schema you can initialize a database. The importer allows feeding the
initial dataset or an updated dataset into the database by collecting data from
a nearby Debian mirror. It is recommended to apply multiarchanalyze.sql once
after the initial data import to avoid lots of changes on unused indices. The
final step is generating the hints by running multiarchanalyze.py. Unless
issuing ANALYZE, the analysis will be slow.