Skip to main content
🔍

Input Arguments

This page outlines the required inputs and optional parameters for running OncoBLADE.

Required Datasets​

1. Signature Matrices​

  • X: An Ngene by Ncell matrix containing average gene expression profiles per cell type (a signature matrix) in log-scale.

  • stdX: An Ngene by Ncell matrix containing standard deviation per gene per cell type (a signature matrix of gene expression variability).

2. Bulk Gene Expression Data​

  • Y: An Ngene by Nsample matrix containing bulk gene expression data. This should be in linear-scale data without log-transformation.

3. Expected Cell Fractions (Optional)​

  • Expectation: An Nsample by Ncell matrix containing the expected cell fractions used to inform OncoBLADE (optional).

Options for Reference RNA-seq Data​

Users have two options for obtaining signature matrices:

  1. Use Pre-processed Signatures

    • Proceed to the Processed Dataset page to download and use pre-processed signature matrices.

  2. Create Signatures from scRNA Data

    • Follow the instructions on the Create Signatures page to generate signatures from single-cell RNA-seq data.

Accepted Input Formats​

Bulk Data​

  • Formats Accepted:

    • CSV: Comma-separated values file.
    • TSV: Tab-separated values file.
    • Excel: .xlsx or .xls files.
    • R Data Files: .RData or .rds files containing expression matrices.

  • Requirements:

    • Rows should represent genes, and columns should represent samples.
    • Gene identifiers should be consistent across all datasets.

Single-cell Data for Signature Creation​

  • Formats Accepted:

    • Seurat Object: RDS files containing Seurat objects with counts and metadata.
    • SingleCellExperiment: RDS files containing SingleCellExperiment objects.
    • Matrix Formats: Sparse or dense matrices with accompanying metadata.
    • Text Files: Counts matrices exported as CSV or TSV files with separate metadata files.

  • Requirements:

    • Cell type annotations must be included in the metadata.
    • Data should be preprocessed (e.g., quality control, normalization) before signature creation.

Model Hyperparameters​

Additional parameters for fine-tuning the OncoBLADE model:

  • Ind_Marker: Index for marker genes.

    • Default: A logical vector [True] * Ngene (all genes used without filtering).
    • Usage: Genes with False are excluded in the first phase (Empirical Bayes) for finding the best hyperparameters.

  • Ind_sample: Index for the samples used in the first phase (Empirical Bayes).

    • Default: A logical vector [True] * Nsample (all samples used).

  • Alpha, Alpha0, Kappa0, sY: Hyperparameters used in the model.

    • Alpha: Optimized during the model run.
    • Defaults:
      • Alpha = 1
      • Alpha0 = 0.1
      • Kappa0 = 1
      • sY = 1

  • IterMax: Maximum number of iterations between variational parameter optimization by L-BFGS and updating hyperparameter Alpha.

    • Default: IterMax = 100

  • Nrep: Number of random initial guesses used to run OncoBLADE.

    • Usage: The best result in terms of the ELBO function will be chosen among the local optima.
    • Default: Nrep = 3

  • Njob: Number of jobs executed in parallel.

    • Default: Njob = 10

Next Steps:

Placeholder Notes: