Journal: Genomics, Proteomics & Bioinformatics

Article Title: TransCell: In Silico Characterization of Genomic Landscape and Cellular Responses by Deep Transfer Learning

doi: 10.1093/gpbjnl/qzad008

Figure Lengend Snippet: An overview of research design A . The number of RNA-seq profiles available per cell line. The numbers were collected from the ARCHS4 website ( https://maayanlab.cloud/archs4/ ). B . Prediction of measurements in six types based on gene expression data of cancer cell lines. The number of cell lines varies across data types. C . Model evaluation process. Due to the high demand for computation power, we started with a small set of measurements for each type and then scaled up to a larger set. D . Schematic of TransCell. The top 5000 features sharing similar distribution between CCLE and TCGA were first selected, followed by the creation of an autoencoder using TCGA pan-cancer tumor transcriptomes. The parameters of the TCGA encoder were then transferred to the second CCLE autoencoder for weight initializations. Afterward, a two-step pre-trained CCLE enc was extracted and linked to a prediction feedforward network. Parameters were tuned automatically (see Method for details). Note that one model is built for each molecular measurement. LASSO, least absolute shrinkage and selection operator; EN, elastic net; RF, random forest; PCA, principal component analysis; DNN, deep neural network; CCLE, Cancer Cell Line Encyclopedia; TCGA, The Cancer Genome Atlas; CCLE enc , CCLE encoder.

Article Snippet: TransCell is composed of two networks: (1) a two-step pre-trained CCLE encoder (CCLE enc ) and (2) a prediction feedforward network (P) ( ).

Techniques: RNA Sequencing, Gene Expression, Selection

Journal: Molecular Cancer

Article Title: Emerging artificial intelligence-driven precision therapies in tumor drug resistance: recent advances, opportunities, and challenges

doi: 10.1186/s12943-025-02321-x

Figure Lengend Snippet: Application of artificial intelligence in basic research on tumor drug resistance

Article Snippet: RNA-seq data from DepMap and TCGA datasets , TransCell (Self-Encoders + Migration Learning + Deep Feedforward Neural Networks) , External validation on proteomic data in CellMinerCDB and RNA-seq data in NCI60 cell line , TransCell improved drug susceptibility prediction performance by more than 50% , [ ] .

Techniques: Biomarker Discovery, Binding Assay, Fluorescence, Kinase Assay, Immunohistochemistry, Quantitative RT-PCR, Staining, Knockdown, CRISPR, Mutagenesis, Knock-Out, Activation Assay, Expressing, Reverse Transcription Polymerase Chain Reaction, Migration, Gene Expression, Plasmid Preparation, Injection, Selection, Histone Deacetylase Assay, Imaging, Cytometry

Journal: Molecular Cancer

Article Title: Emerging artificial intelligence-driven precision therapies in tumor drug resistance: recent advances, opportunities, and challenges

doi: 10.1186/s12943-025-02321-x

Figure Lengend Snippet: Available databases on tumor drug resistance

Article Snippet: RNA-seq data from DepMap and TCGA datasets , TransCell (Self-Encoders + Migration Learning + Deep Feedforward Neural Networks) , External validation on proteomic data in CellMinerCDB and RNA-seq data in NCI60 cell line , TransCell improved drug susceptibility prediction performance by more than 50% , [ ] .

Techniques: Expressing, Mutagenesis, Drug discovery, Biomarker Discovery