MMP Transform Extractor

Category Paths

Follow one of these paths in the Orion user interface, to find the floe.

  • Solution-based/Hit to Lead/Generative Design/Match Molecular Pairs (MMP)

  • Task-based/Virtual Screening - Structure-Based

  • Role-based/Cheminformatician/Medicinal Chemistry Support

  • Role-based/Cheminformatician/Corporate Collection Support

Description

This is the final step of indexing a custom dataset for the Matched Molecular Pair (MMP) method of the Generative Design Floe.

The input is the index (File) generated from MMP indexing Floe. For the transformation extraction, indicate the desired chemistry context for the extracted transforms. The more chemistry context you request, the more targeted the chemistry transformation will be in the generated analogs. Typically separate extraction runs would be performed for each of the desired transformation chemical contexts, with the output stored in the same output collection and then selected during the transformation application activities.

For a ~1M structure index, each run of the Floe extraction run can take approximately 1 hour, and much less for smaller index sets.

Promoted Parameters

Title in user interface (promoted name)

Inputs

MMP Index (mmp_index): The MMP index file to load

  • Required

  • Type: file_in

Outputs

Transform Collection (colname): Name of the output collection

  • Required

  • Type: string

Transform Extract Outputs

Chemistry Context (xform_context):

How many bonds of chemistry context to include in the transformations, larger contexts produce more explicit chemical transformations, smaller context values produce very broad and less specific chemical transformations. The more specific the chemical context specified, the fewer number of examples that may be present as matched molecular pairs from the index.

  • Type: string

  • Default: default

  • Choices: [‘default’, ‘allbonds’, ‘bond0’, ‘bond1’, ‘bond2’, ‘bond3’]

Min MMP count (xform_minmmp): only return transforms having >= this number of Matched Molecular Pairs examples in the index, or 0 to indicate unconstrained

  • Type: integer

  • Default: 6

Max MMP count (xform_maxmmp): only return transforms having <= this number of Matched Molecular Pairs examples in the index, or 0 to indicate unconstrained

  • Type: integer

  • Default: 0

Advanced Options

MMP Example (xform_bestmmp): Optional field to provide an example MMP for each transform

  • Type: string

  • Default: xform_bestmmp

Verbosity (verbosity): Logging verbosity level

  • Type: string

  • Default: warning

  • Choices: [‘info’, ‘warning’, ‘error’, ‘debug’, ‘ddebug’]