schrodinger.application.building_block_exploration.bb_explorer_generate_seed_file module

class schrodinger.application.building_block_exploration.bb_explorer_generate_seed_file.SeedFileGenerator(args)

Bases: object

Class to generate a seed file for building block exploration. The seed file contains compounds synthesized using the provided routes and building blocks. The number of compounds for each route is decided based on what is the total possible number of products that can be synthesized from that route.

If a Bloom filter file is provided, the generated seed file only contains compounds that are in the library represented by the bloom filter. In this case, the user may provide a max_library_attempts parameter to control how many attempts are made to generate one in-library compound. Note that in general, for two step routes, the number of possible products is a lot larger than one-step routes, so the fraction of in-library compounds is likely to be smaller.

__init__(args)

Initialize the SeedFileGenerator with the user-provided arguments. Following attributes of the class are set in this initialization:

: route_dict: Dictionary of synthetic routes to run, user provided or

the default one from mmshare data files.

: log_archive: A zip file containing all the subjob logs.

: state_file: Path to the state file for this job. The state file

contains information about the progress of the job and has all the parameters required to restart the job.

: restart_state_file: Path to the restart state file. If not None,

the workflow assumes that this is a restart job.

: state_dict: Dictionary containing the contents of the state file.

: bb_dir: Directory containing all the building blocks required by

routes we are running. These are either provided by the user or processed from the raw building blocks file provided by the user. If the user does not provide any building blocks, the default mmshare reagent data directory is used.

: product_property_filter: A json file containing property filters

for the product compounds.

: product_smarts_filter: A canvasSearch filter (.cflt) file containing

smarts based filters for the product compounds.

: bloom_filter_path: Path to the bloom filter for in library filtering.

: seed_file: Output seed file that the job returns.

: already_added: Set to keep track of InchiKeys for the compounds

that have been already added to the seed file. This makes sure that we do not add the same compound multiple times.

run()

Main method to setup and run the seed file generation.

setUp()

Setup the seed file generator by: 1) creating the seed file and writing the header line. 2) Preprocessing the building blocks. 3) updating the routes dictionary to only contain routes for which all the required building blocks are present in the bb_dir. 4) adding the seed file and log archive to the job output files.

generateSeedFile()

Generate a seed file based on the provided arguments.

generateSeedCompoundsPerRoute(route_name: str, route_object: RouteNode, route_state: dict)

Generates seed compounds for a specific route. If a bloom filter file is provided, only compounds that are in the library represented by the bloom filter are returned. In that case, initially num_products * max_library_attempts products are synthesized to increase the chances of getting enough in-library compounds. If no bloom filter is provided, num_products * EXTRA_FACTOR_FOR_REPEATED_PRODUCTS products are synthesized to account for repeated products. These products are then filtered through the bloom filter if provided, and upto num_products compounds are added to the seed file.

Parameters:

  • route_name – The name of the route.

  • route_object – The RouteNode object to begin synthesis with.

  • route_state – The state dictionary for this route.

addToSeedFile(product_file: str, route_name: str, num_to_add: int, route_state: dict)

Adds the compounds from the product file to the seed file.

Parameters:

  • product_file – The file containing the products from the synthesis.

  • route_name – The name of the route for which the products were generated.

  • num_to_add – Number of compounds to add to the seed file for this route.

parsePreviousSeedCompounds()

Parses the previously generated seed file in case of a restart and updates the already_added set.