SQL database

A lot of meta-data related to the CC is stored in a PostgreSQL database called chemical_checker_<yyyy>_<mm>. Tables are listed in alphabetical order.

Central tables

The most important tables of the CC are:

• datasets
• It contains dataset specifications.
• Every time a dataset is stored in the CC, this table should be updated.
• general_properties
• Compound properties such as molecular weight or Lipinski's rule of 5 violations.
• ⚠ This table is currently called physchem.
• libraries
• Abbreviation and description of compound libraries.
• ⚠ This table is currently called library_description
• pubchem
• CID, name and synonyms of molecules.
• structure
• It contains inchikey and inchi.
• This is an evergrowing table. Every time a molecule is seen by the CC, it should be stored here.

Analysis tables

Analysis tables, typically related to exemplary datasets.

• coordinate_clust_paired_conditionals
• Conditional probabilities between clusters in different datasets.
• coordinate_conditionals
• Conditional probabilities between similar molecules in different datasets.
• coordinate_correlation
• Canonical correlation analysis between datasets.
• coordinate_paired_conditionals
• Paired conditional probabilities between similar molecules in different datasets.
• coordinate_ranks
• Coincidence of ranks between pairs of similar molecules in different datasets.
• scores
• Popularity, singularity and mappability scores.
• similarity_inference_performance
• Performance analysis of the similarity inference.
• ⚠ This table is currently called mapper_performance.

Auxiliary tables for the production phase

The pipeline requires some

• data
• Mainly used to guide the downloads.
• Whether it is necessary to download again, whether it is a frozen dataset, etc.
• 🤔 I am not sure this table is necessary.
• sims_versions
• This is relevant to the full similarity vectors, especially.
• It tells, for each molecule and coordinate, when was the sims1 calculated and stored in the ./molecules/ directories.
• ⚠ This table is currently called distance_versions.

Auxiliary tables for the CC web app

The CC web app requires some auxiliary tables.

• coordinate_proj_lims
• Contains the number of molecules, and (xlim, ylim) of the projections.
• library_compounds
• Belonging of compounds to exemplary libraries.
• ⚠ This table is currently called libraries
• lookup
• Fast database lookup.
• projections
• (x,y) coordinates of the 2D projections.
• showtargets
• Targets to show in the molecule cards.
• showtargets_description
• Annotations of protein targets.