sigLASSO

"sigLASSO" is a computational tool for analyzing tumor genomes to identify active mutational signatures from the entire repertoire of potential ones. This process is critical for elucidating cancer development mechanisms by optimally decomposing cancer mutation counts, tabulated according to their trinucleotide context, into a linear combination of known signatures. SigLASSO efficiently implements this optimization process through statistical rigor and computational innovation.

Key Features and Functionalities:

- Joint Optimization of Sampling and Signature Fitting: SigLASSO uniquely incorporates the likelihood of multinomial sampling directly into the objective function. This approach is especially beneficial for analyzing datasets where mutation counts are low and sampling variance is high, such as in exome sequencing.

- L1 Regularization for Sparse Solutions: The tool employs L1 regularization to assign mutational signatures to tumor genomes parsimoniously. This regularization technique encourages sparse solutions, making the results easier to interpret and more biologically meaningful.

- Adaptable Model Complexity: SigLASSO can fine-tune its model's complexity based on the data scale and biological priors. This adaptability ensures that the tool can provide the most accurate and informative analysis possible, tailored to the specific characteristics of the dataset.

- Assessment of Model Uncertainty: One of the distinctive features of sigLASSO is its capability to assess model uncertainty, allowing the tool to abstain from making signature assignments in contexts where the confidence level is low, thereby enhancing the reliability of the analysis.

- Sparse and Interpretable Solutions: By leveraging L1 regularization, sigLASSO produces solutions that are not only sparse, minimizing the number of signatures attributed to each sample, but also highly interpretable.

Topic

Oncology;Genetic variation;Exome sequencing;Genomics;Machine learning

Detail

  • Operation: Parsing;Essential dynamics;Standardisation and normalisation

  • Software interface: Command-line interface

  • Language: R

  • License: The GNU General Public License >= v3.0

  • Cost: Free with restrictions

  • Version name: 1.1

  • Credit: The National Institutes of Health and the National Institute of Child Health and Human Development.

  • Input: -

  • Output: -

  • Contact: Mark B. Gerstein mark@gersteinlab.org

  • Collection: -

  • Maturity: -

Publications

  • Using sigLASSO to optimize cancer mutation signatures jointly with sampling likelihood.
  • Li S, et al. Using sigLASSO to optimize cancer mutation signatures jointly with sampling likelihood. Using sigLASSO to optimize cancer mutation signatures jointly with sampling likelihood. 2020; 11:3575. doi: 10.1038/s41467-020-17388-x
  • https://doi.org/10.1038/S41467-020-17388-X
  • PMID: 32681003
  • PMC: PMC7368050

Download and documentation


< Back to DB search