NeuroMDAVIS-FS

Maitra, C., Das, V., Seal, D.B. and De, R.K., 2026. Multi-Omics Integration for Identification of Prognostic Molecular Signatures for Survival Stratification in Lung Cancer. Extention of our earlier work NeuroMDAVIS [1]

Graphical Abstract

Requirements

To run NeuroMDAVIS-FS, one needs to install numpy, pandas, sklearn, scipy, tqdm and tensorflow packages. Installation codes are as follows:

• pip install numpy
• pip install pandas
• pip install scikit-learn
• pip install scipy
• pip install tqdm
• pip install tensorflow

Running NeuroMDAVIS-FS

To run NeuroMDAVIS-FS, import NeuroMDAVIS.py from the utils directory and run the function top_features. All the parameters are mentioned below for better understanding. One can also follow NeuroMDAVIS-FS.ipynb notebook.

Parameters

All input parameters are as follows: X, dim, lambda_act, lambda_weight, num_neuron bs, epoch, sd, and verbose.

• X: List of input data matrices for training. [The input data matrices should be in the form of cells x features.]
• dim: Dimension onto which the data is being projected. default: 2
• lambda_act: Activity regularizer parameter. default: 0
• lambda_weight: Kernel regularizer parameter. default: 0
• num_neuron: Specification of shared and modality specific hidden layers. default: [64, [128, 128, 128]]
• bs: Training batch size. default: 16
• epoch: Total number of iteration for training. default: 500
• sd: To reproduce the results set a seed value. default: 0
• verbose: 1 for printing the training details. default: 1

dim = 2
lambda_act=0.0
lambda_weight=0.0
num_neuron=[64, [128, 128, 128]]
bs=16
epoch=500
sd=1
verbose=1

X = [cnv, rna, pro]
top_feat, model, _, _ = md.top_features(X, dim, lambda_act, lambda_weight, num_neuron, bs, epoch, sd, verbose)

References

---

[1] Maitra C, Seal DB, Das V, De RK. NeuroMDAVIS: Visualization of Single-Cell Multi-Omics Data under Deep Learning Framework. Journal of Computational Biology. 2026;0(0). doi:10.1177/15578666261423965