#071 — Moderated Empirical Bayes t-Test (limma Core)
Statistical Definition
Shrinks gene-wise sample variances toward a common empirical Bayes prior trend across all genes, stabilizing t-statistics in small-n omics.
Bioinformatics Use Case
Microarray, proteomics, metabolomics, and voom-transformed RNA-seq differential expression with small replicate counts (n=3 per group).
Input Variables
Log-normalized expression matrix (G genes x N samples); design matrix and contrast.
Application
In high-throughput proteomics and RNA-seq with 3 to 5 replicates, standard t-tests suffer from noisy variance estimates. Empirical Bayes moderation borrows strength across all genes to rescue true discoveries.
#072 — Gene Set Enrichment Analysis (GSEA) Kolmogorov-Smirnov Permutation Test
Statistical Definition
Computes a weighted running-sum enrichment score along a ranked gene list and assesses significance via phenotype/gene-set permutations.
Bioinformatics Use Case
Detecting subtle, coordinated pathway upregulation across all ranked genes without imposing arbitrary fold-change cutoffs.
Input Variables
Ranked vector of G gene metrics (e.g., signed -log10 p-values); binary membership vector for gene set S.
Application
Individual genes often miss statistical thresholds while an entire metabolic pathway shifts in unison. GSEA captures subtle, coordinated pathway activation across the whole transcriptome.
#073 — Permutational Multivariate ANOVA (PERMANOVA / adonis)
Statistical Definition
Partitions pairwise distance matrices among sources of variation and tests group centroid differences via permutation.
Bioinformatics Use Case
Testing whether gut or tumor microbiome beta-diversity (Bray-Curtis / UniFrac) differs significantly between responders and non-responders.
Input Variables
Pairwise distance matrix D (n x n); grouping factor vector (length n).
Application
Microbiome and single-cell ecosystems exist in high-dimensional distance spaces. PERMANOVA proves whether treatment significantly reshapes overall community composition.
#074 — ANOSIM (Analysis of Similarities)
Statistical Definition
Non-parametric rank-based permutation test comparing within-group versus between-group rank dissimilarities.
Bioinformatics Use Case
Testing community separation across ecological habitats or single-cell clusters from ranked distance matrices.
Input Variables
Pairwise dissimilarity matrix D (n x n); categorical group labels.
Application
ANOSIM converts complex community distances into an intuitive R statistic between 0 and 1, demonstrating how cleanly your biological cohorts separate.
#075 — Tajima's D Neutrality Test
Statistical Definition
Compares normalized pairwise nucleotide differences (pi) against the number of segregating sites (Watterson's theta) to detect evolutionary selection.
Bioinformatics Use Case
Scanning viral or human genomic windows for selective sweeps (D < 0) or balancing selection (D > 0).
Input Variables
Aligned DNA sequences (n haplotypes x L sites).
Application
Identify genomic regions under active evolutionary selection or pathogen drug-resistance sweeps directly from population sequencing alignments.
#076 — McDonald-Kreitman (MK) Selection Test
Statistical Definition
Compares the ratio of nonsynonymous to synonymous polymorphisms within a species against fixed divergence between species using a 2x2 contingency test.
Bioinformatics Use Case
Detecting adaptive protein evolution in rapidly evolving viral proteins or cancer driver genes.
Input Variables
2x2 table: [[Pn (polymorphic nonsyn), Ps (polymorphic syn)], [Dn (fixed nonsyn), Ds (fixed syn)]].
Application
Prove whether a target protein is undergoing positive adaptive evolution rather than neutral genetic drift using the classic McDonald-Kreitman test.
#077 — dN/dS (Ka/Ks) Codon Substitution Likelihood Ratio Test
Statistical Definition
Tests whether the ratio of nonsynonymous to synonymous substitution rates (omega = dN/dS) significantly deviates from neutral evolution (omega = 1).
Bioinformatics Use Case
Identifying positively selected epitope residues in viral spike proteins or drug-binding pockets.
Input Variables
Aligned coding DNA sequences (CDS); neutral null model (omega=1) vs. free-omega model.
Application
Pinpoint exact protein domains under positive selection or strict functional constraint by comparing nonsynonymous versus synonymous codon substitution rates.
#078 — Sequence Kernel Association Test (SKAT)
Statistical Definition
Variance-component score test aggregating bidirectional (protective and deleterious) rare genetic variants within a gene or genomic region.
Bioinformatics Use Case
Whole-exome sequencing (WES) rare-variant gene-burden discovery in complex diseases.
Input Variables
Phenotype vector (y); covariate matrix (X); rare genotype matrix (G of size n x m variants).
Application
Single-variant GWAS misses rare disease mutations. SKAT aggregates rare variants across an entire gene—even when some increase and others decrease risk—to unlock novel drug targets.
#079 — Transmission Disequilibrium Test (TDT)
Statistical Definition
Family-based McNemar-type association test evaluating whether heterozygous parents transmit a risk allele to affected offspring more than 50% of the time.
Bioinformatics Use Case
Trio whole-genome sequencing studies of pediatric neurodevelopmental disorders immune to population stratification.
Input Variables
Counts of transmitted (b) vs. untransmitted (c) risk alleles from heterozygous parents.
Application
Population ancestry differences can create false-positive GWAS hits. Trio TDT uses untransmitted parental alleles as internal controls for 100% ancestry-proof genetic discovery.
#080 — LD Score Regression (LDSC) Genetic Correlation & Polygenicity Test
Statistical Definition
Regresses GWAS chi-square summary statistics against linkage disequilibrium (LD) scores to separate true polygenic signal from confounding stratification.
Bioinformatics Use Case
Estimating SNP-based heritability and genetic correlation between two complex diseases from GWAS summary statistics alone.
Input Variables
Vector of SNP Chi-square statistics; vector of SNP LD scores.
Application
Separate true polygenic disease architecture from population stratification using only GWAS summary statistics—without needing raw individual-level genotypes.
#081 — Colocalization Posterior Probability Test (coloc)
Statistical Definition
Bayesian test computing posterior probabilities (PP.H4) that two traits (e.g., GWAS disease risk and tissue eQTL) share a single causal variant.
Bioinformatics Use Case
Target validation: proving a disease-associated GWAS locus acts by regulating a specific gene's expression in relevant tissue.
Input Variables
GWAS summary stats (beta, varbeta) and eQTL summary stats across M SNPs in a locus.
Application
A GWAS hit near a gene doesn't prove that gene causes the disease. Bayesian colocalization proves whether the disease SNP and gene expression eQTL share the exact same causal variant.
#082 — Mendelian Randomization Inverse-Variance Weighted (IVW) Test
Statistical Definition
Uses genetic variants as instrumental variables to test for a causal effect of a modifiable exposure (e.g., biomarker) on a disease outcome.
Bioinformatics Use Case
Validating whether circulating plasma protein levels causally drive cardiovascular disease or merely reflect secondary inflammation.
Input Variables
Genetic instrument effects on exposure (bx, sex) and outcome (by, sey) across K independent SNPs.
Application
De-risk expensive drug targets before Phase I trials: Mendelian Randomization uses inherited genetic instruments to prove whether your protein target causally drives