A2Sign

A2Sign identifies cell-type-specific molecular signatures from transcriptomic datasets using an agnostic, data-driven approach that does not require prior knowledge of the cell types present.


Key Features:

  • Agnostic Approach: Operates without requiring predefined cell-type information, enabling discovery directly from transcriptomic data.
  • Non-Negative Tensor Factorization (NTF): Decomposes multi-dimensional transcriptomic data into non-negative components corresponding to molecular signatures.
  • Reduction of Collinearities: NTF decomposition reduces collinearities in transcriptomic measurements to clarify biological signals.
  • Inter-Individual Variability: Accounts for variability between individuals to identify signatures that are robust across samples.
  • Universal Application: Applicable to any tissue type, including both healthy and pathogenic contexts.
  • Integration with NMTF Package: Implements non-negative tensor factorization using the NMTF package for computational factorization.

Scientific Applications:

  • Cell-Type Deconvolution: Enables deconvolution of bulk transcriptome data to infer cell-type proportions.
  • Discovery of Novel Signatures: Identifies molecular signatures without prior biological assumptions to support biomarker and interaction discovery.
  • Immune Cell Analysis: Generates molecular signatures for deconvoluting up to 16 immune cell types from microarray and RNA-seq data.

Methodology:

Non-negative tensor factorization (NTF) is applied to multi-dimensional transcriptomic data using the NMTF package.

Topics

Details

License:
MIT
Tool Type:
command-line tool
Programming Languages:
Python
Added:
3/28/2022
Last Updated:
3/28/2022

Operations

Publications

Boldina G, Fogel P, Rocher C, Bettembourg C, Luta G, Augé F. A2Sign: Agnostic Algorithms for Signatures—a universal method for identifying molecular signatures from transcriptomic datasets prior to cell-type deconvolution. Bioinformatics. 2021;38(4):1015-1021. doi:10.1093/bioinformatics/btab773. PMID:34788798.

Links