README.md

pyspaya

Installation

Description

• Spaya API
• Retrosynthesis parameters
• Score

Clients

• REST
• Async
• Callback

Url and Authorization

Installation

pip install pyspaya

Description

Spaya API

This package provides an easy way to score SMILES with the Spaya API.

This restrosynthesis score (RScore) is a metric provided by the Spaya algorithms, related to the probability of a disconnection and consequently to the confidence the algorithm has on this disconnection/route.

Spaya API employs a data-driven AI approach to discover retrosynthetic routes. An iterative exploration of all possible routes is performed until commercially available starting materials are identified.

Useful links:

• Spaya Public
• Spaya API
• Spaya REST documentation
• Spaya Websocket documentation

Score

Each scored SMILES return a RetrosynthesisResult:

from typing import Dict
from iktos.spaya import BearerToken, SpayaClientREST, RetrosynthesisResult

# Create a client
client = SpayaClientREST(url="https://spaya.ai", authorization=BearerToken(token="YourToken"))

# Start a retrosynthesis and wait for the results
result: Dict[str, RetrosynthesisResult] = client.score_smiles(smiles=["c1ccn2nccc2c1"])

# Show results
retro_result: RetrosynthesisResult = result["c1ccn2nccc2c1"]
print(f"Progress:{retro_result.progress} Status:{retro_result.status}")
print(f"Score:{retro_result.rscore} Number of steps:{retro_result.nb_steps}")

Clients

To match every case, this API provides 3 different clients:

• SpayaClientREST: A synchronous client
• SpayaClientAsync: An asynchronous client with reconnection and helpers for the websocket
• SpayaClientCallback: An asynchronous client that allows you to get results using a Callback

REST

SpayaClientREST sends SMILES synchronously to be scored

Example using a list of SMILES:

from iktos.spaya import BearerToken, SpayaClientREST

# Create client with authorization
client = SpayaClientREST(url="https://spaya.ai",
                         authorization=BearerToken("myT0ken"))

# Start a retrosynthesis and wait for the results
scored_smiles = client.score_smiles(smiles=["O=C1CCCCO1", "O=C1CCCNN1",])

# Show the results
for smiles, result in scored_smiles.items():
    print(f"{smiles}: {result.rscore} / {result.nb_steps}")

Example using a DataFrame:

from iktos.spaya import BearerToken, SpayaClientREST
from pandas import DataFrame

# Create client with authorization
client = SpayaClientREST(url="https://spaya.ai/",
                         authorization=BearerToken("myT0ken"))

# Start a retrosynthesis and wait for the results with a dataframe
df = DataFrame({"input_smiles": ["O=C1CCCCO1", "O=C1CCCNN1",]})
df = client.score_smiles(
    smiles=df,
    dataframe_smiles_column="input_smiles",
    callback_progression=lambda p: print(f"progression:{p}")
)

# Show the results
print(f"result: {df}")

Example using a list of SMILES and consume function:

from iktos.spaya import BearerToken, SpayaClientREST

# Create client with authorization
client = SpayaClientREST(url="https://spaya.ai/",
                         authorization=BearerToken("myT0ken"))

# Start a retrosynthesis
client.start_retrosynthesis(smiles=["O=C1CCCCO1", "O=C1CCCNN1",])

# Consume the results as soon as possible
while not client.is_empty:
    for smiles, result in client.consume():
        print(f"{smiles} {result}")

Async

SpayaClientAsync sends SMILES asynchronously to be scored

import asyncio
from iktos.spaya import BearerToken, SpayaClientAsync

async def score():
    async with SpayaClientAsync(url="https://spaya.ai",
                                authorization=BearerToken("myT0ken")
    ) as client:
        # Start scoring SMILES
        await client.start_retrosynthesis(["O=C1CCCCO1", "O=C1CCCNN1",])

        # Wait and print scores as soon as received
        async for smiles, result in client.consume():
            print(f"{smiles}: {result.rscore} / {result.nb_steps}")

asyncio.run(score())

Callback

SpayaClientCallback calls a function as soon as a SMILES is scored

import asyncio
from iktos.spaya import (
    BearerToken, SpayaClientCallback, RetrosynthesisResult
)

async def generator(client: SpayaClientCallback):
    # Generate and start scoring SMILES
    for smiles in ["O=C1CCCCO1", "O=C1CCCNN1"]:
        await client.start_retrosynthesis(smiles)

async def callback(smiles: str, result: RetrosynthesisResult):
    # Handle the results
    print(f"{smiles}: {result.rscore} / {result.nb_steps}")

async def generate_and_score():
    async with SpayaClientCallback(url="https://spaya.ai",
                                   authorization=BearerToken("myT0ken"),
                                   callback=callback) as client:
        # Generate SMILES
        await generator(client)

        # Block until the ends
        await client.wait_result()

asyncio.run(generate_and_score())

Url and Authorization

To access the Spaya API, you need to generate a token from the Spaya API page in your Spaya account

Unless specified, use the BearerToken class

Bearer token

The server is protected by a token

from iktos.spaya import BearerToken, SpayaClientREST
SpayaClientREST(url="https://spaya.ai", authorization=BearerToken(token="itb3cy0s..."))

Custom Bearer token

The server is protected by a token and it uses a custom header

from iktos.spaya import CustomBearerToken, SpayaClientREST
SpayaClientREST(
    url="https://spaya.ai",
    authorization=CustomBearerToken(token="itb3cy0s...", header_key="X-Iktos-Authorization")
)