MKD-CFT

A molecular-knowledge-driven cross-fusion transformer model

Analytics
Example Data:
Click here to load the example data of the MKD-CFT model
Upload Example

Upload Data:
Input Data:
Please upload all Raman and mass spectrometry data of a bacterium along with the corresponding labels, in a format similar to the provided training and test sets. The specific variables are: test_dataset_raman, test_dataset_ms, test_labels_raman and test_labels_ms.
Sources
Train set:
Test set:
Source code:

About the model

The molecular-knowledge-driven cross-fusion transformer model(MKD-CFT) is a multimodal pathogen identification algorithm that investigates the dependencies between mass spectrometry and Raman spectroscopy data to achieve precise pathogen classification. MKD-CFT leverages an innovative two-dimensional data structure and dual-stream architecture. It combines self-attention mechanisms with multi-scale convolutions to capture both local and long-range dependencies, providing a comprehensive understanding of underlying biological phenomena.

Overall workflow