RDKit converter inferring (#4305)

* refactor: move inferring code to separate module

* add docs stub

* refactor as a class

* add template based inferring

* handle deprecate NoImplicit

* fix tests and add missing TemplateInferer tests

* formatting and typing

* remaining formating issues and docs

* add adjust_hydrogens to TemplateInferer

* improve test coverage

* more nitpicks

* add rdDetermineBonds inferer

* address comments and linting

* revert auto-formatting

* expose STANDARDIZATION_REACTIONS and MONATOMIC_CATION_CHARGES for backwards compatibility

* make sanitization optional

* fix rdkit support for numpy v2

* fix tests

* document usage of rdkit inferers

* fix remaining tests rdkit incompatible with numpy 2

* chore: formatting/linting

* address comments: bump to 2.10.0, fix docs, formatting, missing tests

* chore: rename to inferrer

* formatting

* fix failing tests

---------

Co-authored-by: Cédric Bouysset <[email protected]>

Author: cbouy

.github/actions/setup-deps/action.yaml

@@ -75,7 +75,7 @@ inputs:
   pytng:
     default: 'pytng>=0.2.3'
   rdkit:
-    default: 'rdkit>=2020.03.1'
+    default: 'rdkit>=2024.03.4'  # supports both numpy v1 and v2
   scikit-learn:
     default: 'scikit-learn'
   seaborn:

azure-pipelines.yml

@@ -111,7 +111,7 @@ jobs:
       networkx
       parmed
       pytng>=0.2.3
-      rdkit>=2020.03.1
+      rdkit>=2024.03.4
       tidynamics>=1.0.0
     # remove from azure to avoid test hanging #4707
     # "gsd>3.0.0"

package/CHANGELOG

@@ -16,7 +16,7 @@ The rules for this file:
 -------------------------------------------------------------------------------
 ??/??/?? IAlibay, orbeckst, BHM-Bob, TRY-ER, Abdulrahman-PROG, pbuslaev,
          yuxuanzhuang, yuyuan871111, tanishy7777, tulga-rdn, Gareth-elliott,
-         hmacdope, tylerjereddy
+         hmacdope, tylerjereddy, cbouy
 
 
  * 2.10.0
@@ -40,6 +40,8 @@ Fixes
    directly passing them. (Issue #3520, PR #5006)
 
 Enhancements
+ * Added `TemplateInferrer` and `RDKitInferrer` dataclasses to the
+   `RDKitInferring` module to be used by the RDKit converter. (PR #4305)
  * Improve speed of GROMOS11 (TRC) reader (Issue #5079, PR #5080)
  * Improve parsing of topology information from LAMMPS dump files to allow
    reading of mass, charge and element attributes. (Issue #3449, PR #4995)
@@ -63,11 +65,17 @@ Enhancements
    MDAnalysisTest.util (PR #5038)
 
 Changes
+ * Refactored the RDKit converter code to move the inferring code in a separate
+   `RDKitInferring` module. The bond order and charges inferrer has been move to
+   an `MDAnalysisInferrer` dataclass in there. (PR #4305)
  * Removed undocumented and unused attribute
    `analysis.lineardensity.LinearDensity.totalmass` (PR #5007)
  * Remove `default` channel from RTD conda env. (Issue # 5036, PR # 5037)   
 
 Deprecations
+ * The RDKit converter parameter `NoImplicit` has been deprecated in favour of
+   `implicit_hydrogens` and `inferrer` parameters. `max_iter` has been moved
+   under `MDAnalysisInferrer`. (PR #4305)
 
 
 03/11/25 IAlibay, ChiahsinChu, RMeli, tanishy7777, talagayev, tylerjereddy,