Release Notes, Version 2026.2.1

The following products have new versions for this release. Other packages are unchanged from the previous release.

AI Fold Floes Release Notes

v0.2.7 August 2026

General Notice

• This package is built using OpenEye-orionplatform==6.8.3, OpenEye-toolkits==2026.1.0, and OpenEye-Snowball==0.32.0.

Floe Updates

• The output mmCIF file generated by the Boltz prediction cube and saved to the output record now contains _chem_comp_atom and _chem_comp_bond fields for residues and ligands.

• The Protein Sequence to AI Folded Structure Prediction Floe now strictly enforces the input ligand SMILES during Spruce preparation.

• The Protein Sequence to AI Folded Structure Prediction Floe now runs pose checking to compare the Boltz predicted structure to the input ligand SMILES. If a ligand fails a pose check the output structure will need to be reviewed.

• Better optimization has been added for CCD-coded ligand inputs during Spruce preparation.

Classic Lead Optimization Floes Release Notes

v1.0.4 August 2026

General Notice

• This package is built using OpenEye-orionplatform==6.9.0, OpenEye-toolkits==2026.1.0, and OpenEye-Snowball==0.32.4.

Floe Updates

• The BROOD - 3D Fragment Replacement Floe has gone through the following changes:

  • The Query Sketcher now offers three selection options that determine how the floe behaves, chosen directly within the sketcher.

    • Select atoms to keep: the default automatic query generation, where BROOD builds the query from the atoms that are not preserved.

    • Select atoms to replace: the classic selection, matching the floe’s legacy behavior.

    • Select atoms for linking/cyclization: the option to join two or three selected segments through a linking or cyclization search.

  • The floe can now run multiple queries in a single job, supplied either as several prepared BROOD queries or as multiple Query Sketcher instances, provided all queries derive from the same parent molecule.

  • The supported score types now depend on the selection option: Select atoms to keep supports the ROCS and ET score types, while Select atoms to replace and Select atoms for linking/cyclization support all three (ROCS, ET, and LinkOnly).

  • A Build Secondary option has been added and is enabled by default. When enabled, BROOD produces analogs in which several of the input molecule’s query regions are replaced simultaneously, in addition to the analogs where a single region is replaced.

  • Additional BROOD settings are now exposed as parameters, including Maximum Hits, Shape Cutoff, Ring Requirement, and strain-based filtering of the built analogs via Max Strain and Delta Strain.

  • The floe has been redesigned to better accommodate multiple queries and to expose more parallelization and optimization, improving throughput on large jobs. Database reading and the overlay stage are now more parallelized, allowing the floe to make more complete use of the available compute when processing large fragment databases.

  • The default memory requirements have been reduced across the final hit list and report, lowering the floe’s overall resource footprint.

  • The Floe Report has been rewritten with a redesigned, sectioned layout that presents the query depiction, cluster analysis, and per-query fragment hit details. A report is now always produced, and the optional report switch has been removed.

• The BROOD - Query Builder Floe has been updated to incorporate the automatic query creation and linking/cyclization modes described above.

• When an input molecule does not have any atropisomeric bonds, the Calculate Atropisomerism Floe no longer fails trying to depict atropisomers.

Large Scale Floes Release Notes

v5.0.0 August 2026

General Notice

• This package is built using OpenEye-toolkits==2026.1.0 and OpenEye-orionplatform==6.9.0.

Floe Updates

• The Prepare Giga Collections Floe now provides a FastROCS Collection Only switch under the Options parameters, and this is now the default behavior. When the switch is turned off, both FastROCS and Gigadock collections are generated.

• The Prepare Giga Collections Floe supports GPU OMEGA with a variable number of CPUs, allowing more flexible hardware allocation for conformer generation.

• The Prepare Giga Collections Floe uses Thompson sampling for CPU and GPU OMEGA by default with parameter to turn it off under Conformer Generation Settings.

• The FastROCS Plus Floe now supports pocket queries (grid shape queries without molecules). Mixed query types can be provided, including molecule queries, shape queries with molecules, and shape queries without molecules. When a pocket query is present, novelty hit lists are not generated.

• In the FastROCS Plus Floe, custom color force fields are now handled directly within the FastROCS Cube making the use of custom color force fields more computationally efficient.

• A new Attach Query to Results option has been added under Options: Query in the FastROCS Plus Floe. When enabled, the best selected molecule or shape query is written to the output record using the Molecule Query or Shape Query field respectively. When disabled, records are passed through without attaching queries, which is recommended for large queries such as pocket queries.

• The FastROCS Plus and Batch FastROCS Floes now inspect input shape queries and validate that there is no color-atom-type name mismatch between the provided color force field and the shape query.

• The FastROCS Plus Floe now generates the ROCS Hit List independently of the docking step.

• When a design unit is provided and rescoring is disabled in the FastROCS Plus Floe, a ROCS Hit List is not created. The Consensus ROCS Hit List is generated using FastROCS scores as expected.

• The FastROCS Plus Floe now handles mixed query types when using All Queries in the ROCS Rescoring Mode parameter.

• Color atoms are removed from query molecules in the final record to improve overlay visualization.

• The Gigadock and Gigadock Warp Floes now expose a Max Receptor Volume parameter and fail early when a receptor or design unit exceeds this limit. Error messages now include the actual receptor volume to simplify diagnosis.

• The Gigadock Floe now supports selecting a specific design unit or receptor record from the input dataset using the tile view in the Orion UI.

• The Design Unit field name used by the Gigadock Warp Floe has been changed to designunit to align with downstream floes.

• Promoted cube parameters are now unlocked for advanced users.

Protein Modeling Floes Release Notes

v1.1.2 August 2026

General Notice

• This package is built using OpenEye-orionplatform==6.8.0, OpenEye-toolkits==2026.1.0, and OpenEye-Snowball==0.32.0.

Floe Updates

• The Enumerate Pockets on Design Units Floe now promotes site_residue and adds a dedicated Pocket Finding Options parameter group to better control residue-constrained pocket discovery.

ROCS X Floes Release Notes

v1.2.8 August 2026

General Notice

• This package is built using OpenEye-toolkits==2026.1.0, OpenEye-orionplatform==6.9.0, and OpenEye-Snowball==0.32.0.

Floe Updates

• Filtering options (molecular properties, SMARTS, 2D Similarity, etc.) have been added to the ROCS X - Run 3D Search Floe for additional control over the products being searched.

• ROCS X now supports shape queries, including grid shape queries without molecules. These queries can be input into the Multi-Stage ROCS X Search and ROCS X - Initialize 3D Search Floes.

• ROCS X now supports custom color force fields. Custom color force field files and additional color force field options can be input into the ROCS X - Initialize 3D Search and ROCS X - Run 3D Search Floes.

• Thompson sampling for OMEGA is turned on by default in the ROCS X - Run 3D Search Floe.

• The default Number of Molecules to Rescore parameter has been lowered from 100,000 to 50,000 in the ROCS X - Initialize 3D Search Floe.

• The default Product Normalization parameter has been changed to “Set Neutral pH” in the ROCS X - Prepare 3D Library and ROCS X - Run 3D Search Floes.

• ROCS X now fails immediately for macrocycle queries.

Snowball Release Notes

v0.32.5 August 2026

General Notice

• This package can be used with OpenEye-orionplatform>=6.4.0,<7.0.0, OpenEye-toolkits>=2026.1.0, OpenEye-szmap>=2026.1.0, and OpenEye-szybki>=2026.1.0.

New Cubes

• The Spruce Assign Design Unit Radii Cube and Parallel Spruce Assign Design Unit Radii Cube have been added to assign selected radii models to selected components of an input design unit.

• The Get and itemize PDB Codes Source Cube has been added to emit codes, revisions, and UniProt metadata into records for downstream workflows. The cube also allows users to specify UniProt knowledge base target IDs.

• The following new BROOD Cubes have been added to support a redesigned, more parallelized fragment-replacement workflow with support for multiple queries:

  • The Brood Query selection Cube generates query selections from an input molecule and its tagged atom selections, emitting a record for each generated query.

  • The Brood Query Analyzer Cube validates that the incoming queries are generated from the same parent molecule and do not overlap, then emits summary information about the queries.

  • The Brood Overlay Manager and Parallel Brood Overlay Manager Cubes determine which BROOD database packets are relevant to each query and pair them for overlay.

  • The Brood Overlay (Individual) and Parallel Brood Overlay (Individual) Cubes perform BROOD overlay when the query and packet share a record, emitting vectors of OEBroodScores.

  • The Brood Overlay (Series) and Parallel Brood Overlay (Series) Cubes perform BROOD overlay of a single query against a series of database packets, emitting vectors of OEBroodScores.

  • The Brood Scores Hit list Cube builds a ranked hit list from the BROOD scores.

  • The Brood Mol Builder and Parallel Brood Mol Builder Cubes build analog molecules from the candidate BROOD scores (matches).

  • The Brood Secondary Build Manager Cube sets up records for secondary molecule building.

  • The Brood Secondary Molecule Builder and Parallel Brood Secondary Molecule Builder Cubes build secondary analogs that replace multiple query fragments of the original molecule at once.

  • The Brood Hits Hit list Cube builds the final hit list from the built BROOD hits.

Cube Updates

• The Input PDB Codes for Spruce Cube has been enhanced to better normalize mixed old/new PDB code formats (including prefixed and zero-padded forms) and to populate structure revision metadata from the RCSB when not already present.

• The Rotor Bond Count and Parallel Rotor Bond Count Cube Cubes have been updated to use the OEGetRotatableBondCount function to count rotatable bonds.

• The Record to mmCIF/PDB File Converter Cube now supports choosing a record field to drive exported structure filenames.

• The Input PDB Codes for Spruce Cube and Get and itemize PDB Codes Source Cube now use updated RCSB retrieval behavior for structure-supporting data, including support for structure-factor map coefficient sfCIF downloads (2Fo-Fc and Fo-Fc) and updated cryo-EM metadata/map endpoint handling.

• The Create DU from f-pocket pockets and Create DU from OEPocket and f-pocket pockets Cubes now include site-residue-based filtering options for discovered pockets.

• The Spruce Prep, Prepare Biological Units, and Enumerate Sites Cubes now expose toolkit-driven, clashing-solvent delete behavior through cube parameters.

• The Prepare Biological Units and Enumerate Sites Cubes now keep the heavy source record on the master output and store lightweight biological design unit entries in biodu_recvec to avoid duplicating large payloads like cryoEM map files.

• The Parallel Prepare Biological Units Cube now defaults spot_policy to Prohibited.

• The Omega Conformation Cube has been updated to use Thompson sampling by default. The Use Thompson Sampling option can be used to turn off this behavior and revert to exhaustive sampling for torsion driving.

• The Brood Query Cube now builds the BROOD query directly from the atom selection on the input molecule, supporting automatic query generation, classic replacement, and linking/cyclization selections.

• The Brood Report Cube has been rewritten to generate a redesigned, sectioned Floe Report presenting the query depiction, cluster analysis, and per-query fragment hit details.

Removed Cubes

The following structure preparation cubes have been removed from Snowball and are no longer available in this package:

• CreateStructureCollectionCube - Collection management for biomolecular structures

• StructureRecordToShard and ParallelStructureRecordToShard - Structure records to shard conversion

• CreateTempShardBundleRecords - Temporary shard bundling

• BundleTempShardsToShard and ParallelBundleTempShardsToShard - Shard finalization

• StructureRecordsByCodeToShard - Record bundling by code/revision

• CheckForDuplicateCodes - Structure deduplication

Target X Floes Release Notes

v1.0.0 August 2026

General Notice

• This package is built using OpenEye-toolkits==2025.1.1, OpenEye-orionplatform==6.8.3, OpenEye-Snowball==0.31.3, and OpenEye-orionmdcore==2.6.1.

Minor Changes

• The package has been updated to be consistent with Target X branding. This includes modifying the title of the Automated Cryptic Pocket Detection with Probe Occupancy Analysis Floe to Automated Target X Workflow with Probe Occupancy Analysis Floe.

• A fix has been made in the Perform Weighted Ensemble MD Analysis Floe to output correct per-residue RMSF for residues with identical residue numbers across multiple chains in oligomeric proteins.

Utility Floes Release Notes

v2.1.11 August 2026

General Notice

• This package is built using OpenEye-orionplatform==6.8.0, OpenEye-toolkits==2026.1.0, and OpenEye-Snowball==0.32.5.

Floe Updates

• Explicit parameter groupings have been added to the following floes to improve Job Form organization and discoverability: Extract Biological Units, Build Sidechains, Cap Chain Breaks, Swap Metal(s), Mutate Residue(s), Residue State Changer, Rotamers of a Residue, Subset Design Unit, Subset Design Unit to Smallest Binding Unit, Subset Design Unit Within, Superpose DUs, and Update DU Content.

• The Protein Loop Modeling or Re-modeling Floe now exposes a Delete Existing Loop Residues option in the Loop Building Parameter group.

• The DU to mmCIF/PDB File Floe now supports selecting an output title field (Filename Title Field) for exported structure file naming.

• The SPRUCE - Protein Preparation Floe input path has been updated to use the Get and Itemize PDB Codes Cube followed by Input Structure (PDB/MMCIF) Files, enabling improved PDB-code input handling and UniProtKB target-ID-driven code retrieval in the floe input controls.

• The SPRUCE - Protein Preparation Floe input group now includes structure factor sfCIF file input support.

• The OMEGA - 3D Conformer Ensemble Generation Floe now uses Thompson sampling by default for conformer generation. The ROCS-TS mode has been removed from the Conformer Generation Mode option.