Breast cancer–specific CIBERSORTx signature matrices (major, minor, subset levels) and associated reference resources

Title: Breast cancer–specific signature matrices and benchmarking datasets for digital cytometry
Version: 1.0
Date: 2025-12-02
Authors: [Toru Hanamura, Akinori Takase]
Contact: [hanamura.toru.w@tokai.ac.jp]

--------------------------------------------------------------------------------
Overview
--------------------------------------------------------------------------------
This repository contains all datasets generated to build and benchmark the
breast cancer–specific signature matrices described in the manuscript:

"[A multi-resolution, breast cancer–optimized CIBERSORTx reference matrix rigorously validated using single-cell and bulk transcriptomes]"

All data are provided as machine-readable, tab-delimited text files (.txt).
No proprietary formats are used. Each file corresponds exactly to the data
used in the analyses presented in the manuscript and supplementary materials.

These datasets include:
1. Pseudo-bulk gene expression profiles
2. Ground-truth cell-fraction used to generate the pseudo-bulk mixtures (from scRNA-seq)
3. Pure-cell aggregated gene expression profiles (from scRNA-seq)
4. Final signature matrices (major, minor, and subset resolutions)
5. Reference profiles used for S-mode batch correction in CIBERSORTx
6. Source gene-expression profiles used for B-mode batch correction in CIBERSORTx

Each dataset is available at three annotation resolutions:
- **major**
- **minor**
- **subset**

The structure and content of each file type are documented below.

--------------------------------------------------------------------------------
File List and Descriptions
--------------------------------------------------------------------------------

All files are tab-separated (.txt).

====================================
1. PSEUDO-BULK DATA
====================================

### 1-1. Pseudo-bulk expression matrices  
These files contain gene-level pseudo-bulk expression profiles created by
aggregating scRNA-seq data with known cell-type proportions.

- pseudobulk_mixture_major.txt
- pseudobulk_mixture_minor.txt
- pseudobulk_mixture_subset.txt

Columns: Samples  
Rows: Genes (HGNC symbols)  
Values: Normalized expression (non-log TPM)

### 1-2. Ground-truth cell fractions used to generate mixtures  
These matrices provide the true cellular composition of each pseudo-bulk sample.

- pseudobulk_mixture_counts_major.txt
- pseudobulk_mixture_counts_minor.txt
- pseudobulk_mixture_counts_subset.txt

Columns:
  annotation   — Cell-type label (major/minor/subset level)
  cellcount    — Number of single cells aggregated for the mixture
  name         — Pseudo-bulk sample ID (e.g., att_00001)

====================================
2. PURE-CELL AGGREGATED PROFILES
====================================

These files contain aggregated gene expression profiles for each pure cell type
(calculated by averaging all single-cell profiles belonging to each annotation).

- pure_cell_major.txt
- pure_cell_minor.txt
- pure_cell_subset.txt

Columns: Cell types  
Rows: Genes  
Values: Mean expression (non-log TPM)

====================================
3. SIGNATURE MATRICES
====================================

These are the final matrices used for digital cytometry.

- signature_matrix_major.txt
- signature_matrix_minor.txt
- signature_matrix_subset.txt

Columns: Cell types  
Rows: Selected marker genes  
Values: Gene expression used by CIBERSORTx

====================================
4. S-MODE BATCH-CORRECTION REFERENCES
====================================

Reference profiles used as input for CIBERSORTx S-mode correction.

- single_cell_reference_matrix_major.txt
- single_cell_reference_matrix_minor.txt
- single_cell_reference_matrix_subset.txt

====================================
5. SOURCE GENE-EXPRESSION PROFILES
====================================

Gene-expression matrices used as source data before marker selection.

- sourceGEP_major.txt
- sourceGEP_minor.txt
- sourceGEP_subset.txt

--------------------------------------------------------------------------------
How to cite
--------------------------------------------------------------------------------

If you use these datasets, please cite the associated manuscript:

[Full citation once accepted]

For preprints, cite:

[Preprint DOI]

--------------------------------------------------------------------------------
Data format and reproducibility notes
--------------------------------------------------------------------------------

- All files are tab-delimited.
- No missing values are present unless biologically inherent.
- Gene identifiers follow HGNC symbols.
- Samples and cell types use the exact labels consistent with the manuscript.
- Signature matrices are fully compatible with CIBERSORTx (S-mode and B-mode).

--------------------------------------------------------------------------------
License
--------------------------------------------------------------------------------

These datasets are released under the CC BY 4.0 license unless stated otherwise.
Users are free to share and adapt the material with appropriate attribution.

--------------------------------------------------------------------------------
Contact
--------------------------------------------------------------------------------

Department of Breast Oncology
Tokai University School of Medicine
143 Shimokasuya, Isehara-shi, Kanagawa, 259-1193, Japan
E.mail; hanamura.toru.w@tokai.ac.jp