SigmaCCS2: an accurate and robust method for CCS prediction

Official 🤗 interactive demo for the paper "SigmaCCS2: Collision Cross Section Prediction Using Molecular Topology and Geometry Information With Graph Neural Network" and the Github Repo.

Welcome to our online prediction tool for CCS prediction! 😊

Detailed instructions are available for each of the functionality when users click on a tab below 👇.

❗️❗️❗️[Important] Usage tips: Users can use the SigmaCCS2 method to predict the CCS value by directly inputting an adduct type and a SMILES string of a molecule.

Here is a list of the types of adducts that can be predicted:
  • [M+H]+
  • [M+Na]+
  • [M-H]-

We have provided an example of a molecule that can be directly used as input in the Prediction tab.

❗️❗️❗️[Important] How to use:
1️⃣ Upload a CSV file containing the SMILES strings and the types of adducts. Please note that making sure the header column names of the uploaded CSV file are as follows:

2️⃣ Click the Make Predictions button to start predicting CCS values.
3️⃣ View the CCS prediction results online and download the file in CSV format of the entire prediction results! 😊

We have provided an example of the CSV file that can be directly uploaded as input and downloaded by users in the Batch-Predictions tab.

SigmaCCS2 is used to generate an in-silico CCS database for HMDB. The in-silico HMDB database is a collection of calculated m/z and predicted CCS for the three adducts ([M+H]+, [M+Na]+, and [M-H]-) of 217,323 molecules. It can be used for multidimensional filtering to obtain candidate molecules for unknown compounds.

❗️❗️❗️[Important] Usage tips: Users can filter molecules in the HMDB database to obtain candidates within the given thresholds for the unknown compound by inputting its experimental m/z and CCS.

In the case of a compound with m/z of 500.0768 Da and CCS value of 220 Å2, the filtering thresholds for m/z and CCS values are defined as 0.1543 Da and 5%, respectively. This example can be directly used as input in the Query Predicted Database tab.

❗️❗️❗️[Important] How to use:
1️⃣ Upload a CSV file containing the experimental m/z and CCS values of unknown compounds, along with the filtering thresholds for m/z and CCS values. Please note that making sure the header column names of the uploaded CSV file are as follows:

2️⃣ Click the Make Queries button to perform multidimensional filtering to obtain the candidate molecules in the in-silico database within the given thresholds.
3️⃣ View the candidate molecules online and download the file in CSV format of the entire candidate molecules results! 😊

We have provided an example of the CSV file that can be directly uploaded as input and downloaded by users in the Batch-Query Predicted Database tab.

📧 Contact
If you have any questions, please feel free to reach me out at 232303012@csu.edu.cn.