FracFixR 1.1.0

Improvements and New Features

Bug fixes

• Fixed parallel worker setup to use parallelly::availableCores() instead of parallel::detectCores() - 1. This avoids exceeding the hard localhost worker limit enforced by parallelly and respects the mc.cores option set by CRAN/CI environments.

New features

• FracFixR() gains two new parameters st1 (default 0.6) and st2 (default 0.999) that allow the user to control the quantile range used to select informative transcripts for the NNLS regression fit. Previously these thresholds were fixed at 70% and 96%.
• New exported function get_corrected_counts(): converts the proportion matrix from FracFixR() back to an interpretable count matrix by multiplying each sample’s proportions by the Total abundance of the matched replicate.

Internal changes

• TotalSum computation now uses na.rm = TRUE in rowSums() and transcript selection now explicitly filters out NA values, improving robustness when Total samples contain missing data.

FracFixR 1.0.0

Initial CRAN Release

This is the first release of FracFixR, a compositional statistical framework for absolute proportion estimation between fractions in RNA sequencing data.

Features

• Core functionality
  • FracFixR(): Main function for fraction correction using NNLS regression
  • DiffPropTest(): Statistical testing for differential proportions between conditions
  • Support for any RNA fractionation protocol (polysome profiling, subcellular localization, etc.)
• Statistical methods
  • Non-negative least squares (NNLS) regression for fraction weight estimation
  • Three test options: GLM (binomial), Logit, and Beta-binomial Wald
  • Automatic estimation of “lost” unrecoverable fraction
  • FDR correction for multiple testing
• Visualization
  • PlotFractions(): Stacked bar plots of fraction proportions
  • PlotComparison(): Enhanced volcano plots for differential results
  • Diagnostic plots for regression quality
• Performance
  • Automatic parallel processing using available CPU cores
  • Efficient handling of large datasets (tested up to 50,000 transcripts)
  • Memory-efficient implementation

Documentation

• Comprehensive function documentation with examples
• Introductory vignette with workflow demonstration
• Detailed README with troubleshooting guide

Testing

• Unit tests covering all major functions
• Validation against synthetic data with known ground truth
• Reproducibility tests ensuring consistent results

Known Limitations

• Requires at least one “Total” sample per condition-replicate combination
• Minimum of 10 transcripts recommended for reliable regression
• Statistical tests require at least 2 replicates per condition

Future Development

We welcome contributions and feedback. Please report issues at: https://github.com/Arnaroo/FracFixR/issues