efficION is a python program for predicting mass spectrometry relevant compound ionization efficiency. Two interacting deep neural network models are implemented for log ionization efficiency (logIE) value prediction and error attenuation; one sequential model (i.e., model 1) predicts logIE while a second sequential model (i.e., model 2) attempts to correct for residual logIE prediction error produced by model 1.
The program supports single logIE query or batch chemical logIE queries. For single query, either a canonical SMILE or chemical name is applicable, along with a required solution pH value; for batch queries, the following csv file data format is required:
| SMILES | pH | |
|---|---|---|
| 1 | C1=CNC(=O)NC1=O | 7.2 |
| n | ... | ... |
Upon completion of a task a tabulated result similar to the table below is saved to a csv file.
| Chemical Name | SMILE | logIE | |
|---|---|---|---|
| 1 | 1H-pyrimidine-2,4-dione | C1=CNC(=O)NC1=O | 0.49364716 |
| n | ... | ... | ... |
Currently, efficION is only appropriate for predicting logIE relating to the ESI ionization technique.
Google account needed to access Google Colab notebook.
To create a small batch queries csv input file ad hoc:
import pandas as pd
try:
!touch small_batch.csv
except:
pass
column_names=["SMILES","pH"]
small_batch=pd.read_csv("small_batch.csv", names=column_names)
comp_list = #list of compounds -> ["C(=O)=O", "O"]
pH_list = #list of corresponding pH values -> [2.7, 7.2]
small_batch['SMILES'] = comp_list
small_batch['pH'] = pH_list
small_batch.to_csv("small_batch.csv", index=False)
to access the efficION platform.
Liigand, J., Wang, T., Kellogg, J. et al. Quantification for non-targeted LC/MS screening without standard substances. Sci Rep 10, 5808 (2020). https://doi.org/10.1038/s41598-020-62573-z