How to format/proper syntax for REINVENT4 TOML file input?

[IzaCra]

Hi,

I'm very new to the REINVENT4 software, specifically the LinkInvent run mode, and am not completely sure of how I'm supposed to format input files. Currently, I'm trying to run a basic LinkInvent fragment generation to determine if its functionalities will suit my current project. I have been attempting to use the REINVENT UI app https://reinvent-ui.streamlit.app/REINVENT_UI to automatically generate TOML input files, yet when I run REINVENT4 in an Anaconda command prompt with the file, it'll return multiple errors regarding the input. I have ensured that I downloaded the most recent version of REINVENT4, along with the prior model for LinkInvent and the SMILES file. I have pasted the text from the TOML file below. I need your help correcting the file, and if it would be possible, could you please provide a sample TOML LinkInvent input for future reference and guidance? Thank you in advance!

Best,
Izzy

### REINVENT4 TOML input ###
### Reinforcement Learning (RL) Run Mode ###

# General Options
run_type = "staged_learning"
use_cuda = false
tb_logdir = "TensorBoard_RL"
json_out_config = "RL_inputWarheadsTest.json"

[parameters]
# RL Parameters
summary_csv_prefix = "summary_RLWarheadsTest"
use_checkpoint = false
batch_size = 150
unique_sequences = true
randomize_smiles = true

# LinkInvent Molecule Generator
prior_file = "linkinvent.prior"      #Full file directory is used in TOML input
agent_file = "linkinvent.prior"     #Full file directory is used in TOML input
smiles_file = "warheads.smi"

# Learning Strategy Parameters
[learning_strategy]
type = "dap"
sigma = 1
rate = 0.0001

# Diversity Filter Parameters
[diversity_filter]
type = "IdenticalMurckoScaffold"
bucket_size = 25
minscore = 1.0

# Stage Parameters
[[stage]]
chkpt_file = "RL_calcWarheadsTest.chkpt"
termination = "simple"
max_score = 1.00
min_steps = 10
max_steps = 100

#Scoring Components Parameters
[scoring]
type = "geometric_mean"
parallel = 1

[[scoring.component]]
[scoring.component.NumRotBond]
[scoring.component.NumRotBond.endpoint]
name = "NumRotBond"
weight = 1.0
min = 0
max = 5
transform.type = "Step"
transform.high = 5.0
transform.low = 0.0

[[scoring.component]]
[scoring.component.MolecularWeight]
[scoring.component.MolecularWeight.endpoint]
name = "MolecularWeight"
weight = 1.0
min = 50
max = 500
transform.type = "Sigmoid"
transform.low = 50.0
transform.high = 500.0
transform.k = 0.5

[halx]

Hi Izzy,

many thanks for your feedback. We have example files in the support/ directory which should give you a start. Without seeing the error message I will not be able to diagnose your problem.

Cheers,
Hannes.

[IzaCra]

Hi Hannes,
Thank you for your reply! I apologize, I should have mentioned the type of error earlier. I'm mainly receiving a validation error, which I have attached below, with the username and beginning of the file paths omitted. Once again, thank you for your help.

Best,
Izzy

Traceback (most recent call last):
  File "conda\envs\reinvent4iza\lib\runpy.py", line 196, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "conda\envs\reinvent4iza\lib\runpy.py", line 86, in _run_code
    exec(code, run_globals)
  File "conda\envs\reinvent4iza\Scripts\reinvent.exe\__main__.py", line 6, in <module>
  File "REINVENT4\reinvent\Reinvent.py", line 196, in main_script
    main(args)
  File "REINVENT4\reinvent\Reinvent.py", line 165, in main
    runner(
  File "REINVENT4\reinvent\runmodes\RL\run_staged_learning.py", line 43, in run_staged_learning
    config = RLConfig(**input_config)
  File "conda\envs\reinvent4iza\lib\site-packages\pydantic\main.py", line 250, in __init__
    validated_self = self.__pydantic_validator__.validate_python(data, self_instance=self)
pydantic_core._pydantic_core.ValidationError: 1 validation error for RLConfig
scoring
  Extra inputs are not permitted [type=extra_forbidden, input_value={'type': 'geometric_mean'...': 500.0, 'k': 0.5}}}}]}, input_type=dict]
    For further information visit https://errors.pydantic.dev/2.12/v/extra_forbidden

[halx]

The error message tells you that you need to prefix all scoring with stage.. Otherwise the TOML parser does not know that those sections are subsections of scoring. So e.g.

[stage.scoring]
type = "geometric_mean"
parallel = 1

[[stage.scoring.component]]

etc.