Bioconductor Genefilter

genefilter

Dependencies & Environment

Package-intrinsic requirements from the Bioconductor landing page — reproduce in any R environment.

• Version: 1.94.0 · Bioconductor: 3.23 · R: ≥ 4.6
• Depends: AnnotationDbi, annotate, Biobase, survival
• Imports: MatrixGenerics, AnnotationDbi, annotate, Biobase, survival
• System requirements: URL
• Install: BiocManager::install("genefilter")

When to Use

• Filtering genes from a microarray or expression dataset according to specific or non-specific filtering mechanisms using genefilter.
• Selecting genes that have an expression measure above a certain threshold in at least a minimum number of samples using kOverA.
• Finding genes that are close to specific genes of interest based on distance measures using genefinder.
• Performing independent filtering to adjust p-values and increase power using filtered_p and filtered_R.

When NOT to Use

• For modern RNA-seq count-based differential expression; use DESeq2 or edgeR built-in filtering because they are optimized for negative binomial distributions.
• For single-cell RNA-seq data; use scran or Seurat because they handle high sparsity and dropout rates better than basic variance/mean filters.

Data Requirements

• Input expression data typically provided as an ExpressionSet object (e.g., from the Biobase package) or a numeric matrix.
• Covariates or sample metadata (e.g., factors with two levels) for specific filtering like ttest.

Key Parameters

• k (5): The minimum number of samples required to exceed the threshold in kOverA.
• A (200): The expression measure threshold in kOverA.
• method ("euc"): The distance measure used in genefinder (e.g., "euc", "maximum", "manhattan").
• scale ("none"): Controls the scaling of the rows in genefinder (e.g., "none", "range", "zscore").
• theta (0.5): The filtering fraction/threshold used in filtered_p and filtered_R.
• p (0.1): The p-value threshold used in ttest for specific filtering.

Best Practices

• Assemble individual filtering criteria (like kOverA or ttest) into a combined filtering function using filterfun before applying it with genefilter.
• Use rowSds, rowVars, or rowttests for fast row-wise statistical calculations on expression matrices.
• Visualize the effect of filtering on multiple testing adjustments using rejection_plot or filter_volcano.

Common Pitfalls

• Biased multiple testing: Using a specific filter (e.g., differential expression p-values) before applying multiple testing correction. Fix: Use non-specific filters (like overall variance via rowVars) for independent filtering.
• Scale dominance in distance metrics: Finding nearest genes with genefinder without scaling can be dominated by overall expression magnitude. Fix: Set scale="zscore" or scale="range" in genefinder to normalize row variances.
• Slow row-wise operations: Using apply with standard t.test or var on large matrices. Fix: Use the optimized rowttests, rowSds, and rowVars functions provided by the package.

Alternatives

• DESeq2: For RNA-seq independent filtering integrated directly into the results extraction.
• edgeR: Provides filterByExpr which is specifically designed for count-based library size adjustments.
• matrixStats: For fast row/column-wise statistics if filtering functions are not needed.

Citations

• Bourgon R, Gentleman R, Huber W. (2010). "Independent filtering increases power for detecting differentially expressed genes." Proceedings of the National Academy of Sciences.

References

• Homepage: bioconductor.org/packages/genefilter
• Vignette: vignette_0_5fb6bf3b.txt

Run this on BioMate

This skill is the knowledge layer — when, why, and how to use genefilter. To run this analysis on your own data with managed compute, automated QC, and reproducible outputs, use BioMate — free to start.

▶ Open genefilter on BioMate →