Synthesis Module

The synkit.Synthesis package offers tools for both reaction prediction and chemical reaction network (CRN) exploration.

Reaction Prediction: Reactor 

The synkit.Synthesis.Reactor submodule supports two backends:

NetworkX-based reactor SynReactor
MØD-based reactor MODReactor [3].

Reactor parameters

Name	Type	Default	Description
`invert`	bool	`False`	Use `False` for forward prediction (substrate → products); `True` for backward prediction (target → precursors).
`explicit_h`	bool	`False`	When `True`, all hydrogens within the reaction center are rendered explicitly in the resulting SMARTS.
`strategy`	str	`'bt'`	Graph-matching strategy to enumerate transformations: - `'comp'`: component-aware subgraph search (fastest, preferred) - `'all'`: exhaustive arbitrary subgraph search - `'bt'`: backtracking fallback (tries `comp` first, then `all` if no match)

Example: Forward Prediction

Forward reaction prediction with explicit H and backtracking strategy

from synkit.Synthesis.Reactor.syn_reactor import SynReactor

input_fw = 'CC=O.CC=O'
template = '[C:2]=[O:3].[C:4]([H:7])[H:8]>>[C:2]=[C:4].[O:3]([H:7])[H:8]'

reactor = SynReactor(
    substrate=input_fw,
    template=template,
    invert=False,        # forward prediction
    explicit_h=True,     # show H in reaction center
    strategy='bt'        # try component search, then exhaustive
)

smarts_list = reactor.smarts_list
print(smarts_list)
# >> ['[CH3:1][CH:2]=[O:3].[CH:4]([CH:5]=[O:6])([H:7])[H:8]>>[CH3:1][CH:2]=[CH:4][CH:5]=[O:6].[O:3]([H:7])[H:8]', '[CH3:4][CH:5]=[O:6].[CH:1]([CH:2]=[O:3])([H:7])[H:8]>>[CH:1]([CH:2]=[O:3])=[CH:5][CH3:4].[O:6]([H:7])[H:8]']

Example: Backward Prediction

Backward reaction prediction targeting product to precursors

from synkit.Synthesis.Reactor.syn_reactor import SynReactor

target = 'CC=CC=O.O'
template = '[C:2]=[O:3].[C:4]([H:7])[H:8]>>[C:2]=[C:4].[O:3]([H:7])[H:8]'

reactor_bw = SynReactor(
    substrate=target,
    template=template,
    invert=True,         # backward prediction
    explicit_h=False,    # hydrogens implicit
    strategy='comp'      # component-aware search
)

precursors = reactor_bw.smarts_list
print(precursors)
# >> ['[CH3:1][CH:2]=[O:6].[CH3:3][CH:4]=[O:5]>>[CH3:1][CH:2]=[CH:3][CH:4]=[O:5].[OH2:6]', '[CH3:1][CH3:2].[CH:3]([CH:4]=[O:5])=[O:6]>>[CH3:1][CH:2]=[CH:3][CH:4]=[O:5].[OH2:6]']

Example: Implicit Hydrogen (NetworkX)

In some cases you may want to use an implicit-H template. Set implicit_temp=True and explicit_h=False:

Implicit-H template for backward prediction

from synkit.Synthesis.Reactor.syn_reactor import SynReactor

target = 'CC=CC=O.O'
template = '[C:2]=[O:3].[CH2:4]>>[C:2]=[C:4].[OH2:3]'

reactor_imp = SynReactor(
    substrate=target,
    template=template,
    invert=True,          # backward prediction
    explicit_h=False,     # hydrogens implicit
    strategy='comp',      # component-aware search
    implicit_temp=True    # use implicit-H template
)

precursors = reactor_imp.smarts_list
print(precursors)
# >> ['[CH3:1][CH:2]=[O:6].[CH3:3][CH:4]=[O:5]>>[CH3:1][CH:2]=[CH:3][CH:4]=[O:5].[OH2:6]', '[CH3:1][CH3:2].[CH:3]([CH:4]=[O:5])=[O:6]>>[CH3:1][CH:2]=[CH:3][CH:4]=[O:5].[OH2:6]']

Example: Forward Prediction (MØD)

Forward prediction (no atom-map reservation)

from synkit.Synthesis.Reactor.mod_reactor import MODReactor

input_fw = 'CC=O.CC=O'
template = '[C:2]=[O:3].[C:4]([H:7])[H:8]>>[C:2]=[C:4].[O:3]([H:7])[H:8]'

reactor_mod = MODReactor(
    substrate=input_fw,
    rule_file=template,
    invert=False,         # forward direction
    strategy='bt'         # component-aware, then exhaustive
)

reaction_list = reactor_mod.reaction_smiles
print(reaction_list)
# >> ['CC=O.CC=O>>CC=CC=O.O']

Example: Backward Prediction with AAM (MØD)

Backward prediction (with atom-map preservation)

from synkit.Synthesis.Reactor.mod_reactor import MODReactor
from synkit.Synthesis.Reactor.mod_aam      import MODAAM
from synkit.IO                            import smart_to_gml

input_bw = 'CC=CC=O.O'
rule_gml = smart_to_gml(
    '[C:2]=[O:3].[C:4]([H:7])[H:8]>>[C:2]=[C:4].[O:3]([H:7])[H:8]',
    core=True
)

reactor_bw = MODAAM(
    substrate=input_bw,
    rule_file=rule_gml,
    invert=True,
    strategy='bt'
)

smarts_list = reactor_bw.get_smarts()
print(smarts_list)
# >> [
#   '[CH3:1][CH:2]=[O:3].[CH:4]([CH:5]=[O:6])([H:7])[H:8]>>[CH3:1][CH:2]=[CH:4][CH:5]=[O:6].[O:3]([H:7])[H:8]',
#   '[CH3:1][CH:2]([H:3])[H:4].[CH:5]([CH:6]=[O:7])=[O:8]>>[CH3:1][CH:2]=[CH:5][CH:6]=[O:7].[H:3][O:8][H:4]'
# ]

Chemical Reaction Networks: CRN 

The synkit.Synthesis.CRN submodule provides classes for constructing, analyzing, and exploring chemical reaction networks derived from reaction rules.

class synkit.Synthesis.CRN.mod_crn.MODCRN(rule_db_path: str | List[str], initial_smiles: List[str], n_repeats: int = 2)

Bases: object

MODCRN ======

High-level class for constructing, inspecting, and reporting a chemical reaction network using the MØD derivation graph (DG) API.

Key Features

Flexible rule loading: accept a JSON database path or a list of GML strings.
Initial molecule deduplication via graph isomorphism.
Customizable iterative strategy for rule applications.
Built-in summary and external report export.

Parameters

rule_db_pathUnion[str, List[str]]: Path to a JSON rule database or list of GML rule strings.
initial_smilesList[str]: SMILES strings of the seed molecules.
repeatsint, default=2: Number of repeat cycles for rule application.

Properties

rulesList[Rule]: Loaded Rule objects for network construction.
graphsList[Graph]: Deduplicated initial Graph objects.
derivation_graphDG: The active derivation graph instance.
num_verticesint: Number of molecules in the derivation graph.
num_edgesint: Number of reactions in the derivation graph.

Methods

build() -> None: Populate the derivation graph with the configured strategy.
print_summary() -> None: Print and save a concise summary of the derivation graph.
export_report(path: str) -> None: Generate an external report via the mod_post CLI.
help() -> None: Print usage examples and API summarylog.

build() → None: Populate the derivation graph using: addSubset(initial) >> repeat[repeats](rules)

property derivation_graph: None: The active derivation graph instance.

export_report() → None: Generate an external report via the mod_post CLI.

property graphs: List[None]: Deduplicated initial Graph objects.

help() → None: Print usage examples and API summary for MODCRN.

property num_edges: int: Number of reactions in the derivation graph.

property num_vertices: int: Number of molecules in the derivation graph.

print_summary() → None: Print and save a concise summary of the derivation graph.

property rules: List[None]: Loaded Rule objects for network construction.