TransCell is a web portal that performs predictions of molecular features and cellular responses, including metabolite, protein, gene effect score, copy number variation, drug sensitivity and mutation based on more than 350,000 models built using gene expression. TransCell took advantage of the transfer learning technique utilizing source pan-cancer tumor samples to learn a better weight initialization for downstream target task on cell lines.

  • To use TransCell simply upload gene_expression data file in csv format - genes as rows and samples as columns. The uploaded gene_expression.csv file should be in log2-transformed tpm scale. Human Gene symbols or ENSEMBL gene IDs should be under 'genes' (case-sensitive) column.

  • TransCell currently supports six samples at a time. If the uploaded gene_expression.csv file does not contain our feature genes,those genes will be replaced by the median of their corresponding genes in our original database.

  • We only provide 'All' feature predictions for metabolites and proteins. For the rest of the four types of predictions,you need to select a maximum 20 features from the feature list.

  • Depending upon the number of samples and features,the job running time varies between 5-15 minutes. The result file can be downloaded from the portal or will be sent via email if a valid email ID is provided.

  • Brief information about 6 molecular features can be seen here.

  • Step by step instructions to use TransCell with this sample file.

Upload gene_expression csv file: *
Select molecular type for prediction:
Predict Features:
Select features from list (Max 20): *
Job title: *
Email Address (optional):
Download Results
To cite TransCell: Shan-Ju Yeh, et al, TransCell: In silico characterization of genomic landscape and cellular responses from gene expressions through a two-step deep transfer learning, 2023,Genomics, Proteomics & Bioinformatics, qzad008 biorxiv.
TransCell code is available at Github. The portal is developed by the Chen Lab. Contact: for any questions. @copyright 2022.All rights reserved.