ROCS X - Hit List Clustering and Sampling
Description
This floe performs several functions for triaging and showcasing diversity in a hit list from a search. First, it reads in a hit list and generates Bemis-Murcko scaffolds (regular and uncolored) for each hit. Then it clusters the hits based on 2D fingerprint similarity using Sphere Exclusion Clustering. The floe returns a subset of the best hits from the best clusters (defaults are up to 5 hits from the top 25 clusters), in addition to the full clustered hit list for further analysis.
Key Inputs and Outputs
The key input is a hit list dataset, which is typically output from the ROCS X - Run 3D Search Floe. Molecules to be used as known cluster heads can be provided as optional input (typically the ROCS X search query molecule is provided as a known cluster head).
The key outputs are a subsampled hit list, which is a subset of the best hits from top clusters, and a clustered hit list, which is the full hit list including cluster assignments and Bemis-Murcko scaffolds.
Promoted Parameters
Title in user interface (promoted name)
Inputs
Hit List (in): Hit list dataset to cluster and sample.
Type: data_source
Hit List File (files_in): Hit list file to cluster and sample.
Type: file_in
Known Cluster Heads (known_heads): (Optional) Dataset of known cluster heads to include in clustering.
Type: data_source
Missing Sort Value (Numeric) (missing_sort_value): If set, records missing the sort value will have this value for sorting instead of being discarded. If an integer field is used for sorting, this value will be cast to integer. Useful for ensuring records with missing sort values are retained.
Type: decimal
Default: 2.0
Outputs
Sampled Hit List Output (sampled_out): Output dataset for the sampled hit list.
Required
Type: dataset_out
Default: sampled_hitlist.oedb
Full Clustered Hit List Output (clustered_out): Output dataset for the full clustered hit list.
Required
Type: dataset_out
Default: clustered_hitlist.oedb
Clustering Settings
Similarity Threshold (similarity_cutoff): The similarity threshold used for exclusion clustering. Molecules with similarities above this threshold are considered members of the same cluster.
Type: decimal
Default: 0.6
Fingerprint Type (fp_type): The fingerprint type generated for similarity calculation.
Type: string
Default: Circular
Choices: [‘Circular’, ‘Lingo’, ‘MACCS’, ‘Path’, ‘Tree’, ‘CircularVS’, ‘PathVS’, ‘TreeVS’]
Subsampling Settings
Cluster Label Field (label_field): Integer field to use as label for grouping records. Overridden by string field if both are set.
Type: field_parameter::int
Default: Cluster ID
Sort Field (sort_field): The field name used to sort records within each label group. Overrides by int field if both are set.
Type: field_parameter::float
Default: Tanimoto Combo
Sort Order (sort_order): The sort order for the scores. Descending means higher scores are better, Ascending means lower scores are better.
Type: string
Default: Descending
Choices: [‘Descending’, ‘Ascending’]
Rank Field (rank_field): The field name used to provide the final rank of the sorting. If left unset, the rank will not be placed onto the record.
Type: field_parameter::int
Default: Hit List Rank
Number of Samples per Cluster (num_samples_per_cluster): The number of molecules to sample from each cluster.
Type: integer
Default: 5
Number of Clusters to Sample (num_clusters_to_sample): The number of clusters from which to sample molecules. Set to 0 to sample all clusters.
Type: integer
Default: 25
Bemis-Murcko Settings
Hetero Bemis-Murcko Scaffold SMILES Field (bm_scaffold_smiles_field): The name for the SMILES field.
Required
Type: field_parameter::string
Default: Hetero Bemis Murcko
Bemis-Murcko Scaffold SMILES Field (uncolored_bm_scaffold_smiles_field): The name for the SMILES field.
Required
Type: field_parameter::string
Default: Bemis Murcko
Bemis-Murcko Framework Field (bm_framework_field):
Required
Type: field_parameter::mol
Default: Hetero BM Scaffold Mol
Uncolored Bemis-Murcko Framework Field (uncolored_bm_framework_field):
Required
Type: field_parameter::mol
Default: BM Scaffold Mol
Advanced Settings
Disk Space (MiB) (disk_space): The amount of disk space required for the main cube group. For very large hit lists or hit lists containing records with large amounts of attached data, this may need to be increased.
Type: decimal
Default: 5120.0
Known Heads Disk Space (MiB) (kh_disk_space): The amount of disk space required for the known heads cube group. For an extremely large number of known cluster heads, this may need to be increased.
Type: decimal
Default: 5120.0