The biologically informed classifier uses the graph convolutional neural network to do multi-label classification for hierarchical tumor subtypes with tiers.

The tool automates biomedical literature analysis from PubMed. It builds networks of gene and concept co-occurrences from user-selected publications, expands research through new article discovery, and identifies relevant interactions, including novel ones. 

The pipeline allows users to perform end-to-end analysis of bulk miRNA sequencing data, including quality control of FastQ files, mapping of read counts to miRNA genes using miRBase or Reference genome, quantification of miRNA read counts, differential gene expression analysis using DEseq2, gene set enrichment analysis using curated cancer hallmark gene sets, and identification of miRNA targets.

We characterized miRNA-gene interactions in over 2000 REACTOME pathways in dendritic cells used for immunotherapy. All the identified interactions are visualized in complex heat maps accompanied with a table for detailed information for further research.

The database contains cooperative miRNA pairs that are predicted for efficiently modulating the expression of protein-coding genes of cancer driver genes underlying metastatic melanoma.

The pipeline builds a deep learning model based on disease networks. The learned representations are used for patient stratification, and  in turn, the data mined in the network provides biomedical explanation to the deep learning model.

The pipeline presents a network-based method for identifying cancer genes that can be used as prognostic and prognostic biomarkers and a advanced drug repurposing method targeting those cancer.

The multi-level mathematical model accounts for the anti-cancer effects of an immune vaccine at the molecular level and can be used for optimizing the vaccine design.