BARNO

Batch-Aware Regulatory Network Optimization

BARNO is an R package for batch-aware optimization of transcription factor regulatory weights in single-cell transcriptomic analysis. The framework is designed to reduce sample-group-specific or batch-driven transcription factor signals while preserving biologically coherent regulatory structure.

Author E-mail

zhang.xi6@northeastern.edu

Preprint

The BARNO framework is described in our preprint:

BARNO: a batch-aware regulatory network optimization framework reveals a RAN-ENO1-NONO core in melanoma

DOI: https://doi.org/10.21203/rs.3.rs-9142292/v1

Run Example & Data Archive

DOI: https://doi.org/10.5281/zenodo.19301264

Pipeline

Overview

BARNO builds on GENIE3-based regulatory network inference and refines transcription factor–target gene weights by integrating:

• regulatory importance from GENIE3
• transcription factor–target gene correlation
• temporal consistency inferred from transcriptional scoring
• batch-aware penalization based on spatial dispersion in PCA space

Using this framework, BARNO supports:

• construction of regulatory weight matrices
• batch-aware regulator prioritization
• survival-associated module selection
• core regulator identification
• layered gene regulatory network (GRN) reconstruction

Repository structure

BARNO/
├── R/            # Core package functions
├── Run/          # Example analysis scripts
├── man/          # Function documentation
├── figures/      # Figures used in the manuscript / README
├── data/         # Data notes and small example files
├── DESCRIPTION
├── NAMESPACE
└── README.md

Installation

You can install BARNO from GitHub using devtools:

install.packages("devtools")
devtools::install_github("xxxxxi0001/BARNO")

Then load the package:

library(BARNO)

Data Availability

The full datasets used in this study are too large to be distributed directly in this repository.

Original melanoma single-cell RNA-seq data analyzed in this study are available from the Broad Institute Single Cell Portal under accession SCP109: https://singlecell.broadinstitute.org/single_cell/study/SCP109/melanoma-immunotherapy-resistance

Users should download the original dataset separately and provide the local file path when running the BARNO pipeline.

• For researcher using this package for their own study, please make sure your data has following features in RDS file:

  1. tpm
  2. treated: labeled with TRUE/FALSE
  3. cells

Example Workflow (for this study)

1. Load Weight

1). with Batch Effect

set.seed(123)
melanocytic_genes <- c("TYR", "DCT", "MLANA", "PMEL", "MITF", "SOX10", "MIA", "GPNMB", "S100B", "ERBB3")
weight_matrix_melanocytic<-TR_weight("Mel.malignant.rds", # parameter RDS_file
                                      melanocytic_genes, # parameter gene_list
                                      genie_file = "genie_matrix.rds") # parameter genie_file, if NULL, run GENIE3 automatically

2). without Batch Effect

set.seed(123)
weight_matrix_melanocytic<-TR_weight("Mel.malignant.rds", # parameter RDS_file
                                      batch_effect=FALSE, # has to set this parameter to FALSE, cause default TRUE
                                      genie_file = "genie_matrix.rds") # parameter genie_file, if NULL, run GENIE3 automatically

2. Load TF

Melanocytic_TRW<-readRDS("TR_Weight_Melanocytic_punished.rds")
top_genes_Melanocytic<-top_TRweight_genes(Melanocytic_TRW)
saveRDS(top_genes_Melanocytic,file="Top_TF_of_Melanocytic_punished.rds")
top_tf_Melanocytic<-readRDS("Top_TF_of_Melanocytic_punished.rds")

3. Print out TF

head(top_tf_Melanocytic$top_positive_genes,20)
head(top_tf_Melanocytic$top_negative_genes,20)
head(top_tf_Melanocytic$top_positive_summary,20)
head(top_tf_Melanocytic$top_negative_summary,20)

4. Cox Regression Construction & Choose Important TF
5. Run KM

weight_matrix_melanocytic<-readRDS("TR_Weight_Melanocytic_punished.rds")
df_cof_melanocytic_STAT1<-select_co_factor(weight_matrix_melanocytic,"STAT1")
signature_genes<-head(df_cof_melanocytic_STAT1$Gene,10)
melanocytic_STAT1_fit<-km_curve(signature_genes,"TCGA_SKCM.rds","STAT1")

6. Run GRN

weight_matrix_melanocytic<-readRDS("TR_Weight_Melanocytic_punished.rds")
mst_melanocytic_STAT1<-layered_graph("STAT1",weight_matrix_melanocytic)
directed_mst_visualization(mst_melanocytic_STAT1,"STAT1",aspect_ratio = 0.5,text_size=0.6,text_centrality=0,arrow_size=1)

Example Penalty Score Generation

penalty_df<-generate_penalty_score(genie_matrix,df_tpm)