Description

This curriculum spans the full lifecycle of microarray analysis, equivalent in depth to a multi-phase bioinformatics project involving experimental planning, regulatory-grade data processing, cross-platform integration, and stakeholder-specific reporting within a research-intensive organisation.

Module 1: Experimental Design and Sample Selection

Determine appropriate sample size using power analysis to detect biologically relevant expression differences while accounting for expected variability in tissue sources.
Select matched controls and cases to minimize confounding variables such as age, gender, and comorbidities in clinical studies.
Decide between one-color and two-color microarray platforms based on throughput needs, cost constraints, and availability of reference RNA.
Implement randomization of sample processing order to reduce batch effects during hybridization and scanning.
Establish criteria for sample exclusion due to poor RNA quality (e.g., RIN < 7) prior to array hybridization.
Coordinate with clinical teams to ensure ethical compliance and proper annotation of patient-derived samples.
Balance biological replicates versus technical replicates based on project budget and statistical requirements.

Module 2: Microarray Platform Selection and Data Acquisition

Evaluate probe specificity and genome coverage when selecting between Affymetrix, Agilent, and Illumina platforms for a given organism.
Negotiate access to proprietary array formats and ensure compatibility with institutional core facility equipment.
Configure scanner settings (e.g., PMT voltage) to maximize signal dynamic range without saturation.
Implement standardized protocols for RNA labeling, fragmentation, and hybridization to ensure reproducibility.
Monitor hybridization efficiency using spike-in controls and assess spatial artifacts on raw image files.
Validate probe performance by checking for cross-hybridization risks using in silico alignment tools.
Document all instrument settings and reagent lots for audit and replication purposes.

Module 3: Preprocessing and Quality Control

Apply background correction methods (e.g., RMA, MAS5) based on array type and noise distribution characteristics.
Identify outlier arrays using PCA plots, density distributions, and hierarchical clustering of raw intensities.
Correct for spatial artifacts and grid misalignment during image gridding using platform-specific software.
Implement quantile normalization for one-color arrays while preserving biological variation across samples.
Assess RNA degradation effects by analyzing 3’/5’ probe intensity ratios for housekeeping genes.
Filter low-intensity probes that fall below detection thresholds across multiple samples.
Generate standardized QC reports using Bioconductor packages (e.g., arrayQualityMetrics) for team review.

Module 4: Normalization and Batch Effect Correction

Choose between global and intensity-dependent normalization methods based on MA plot asymmetry.
Detect batch effects using surrogate variable analysis (SVA) when samples are processed in different labs or time points.
Apply ComBat to adjust for known batches while preserving biological signal in differential expression analysis.
Validate correction efficacy by checking cluster separation in PCA before and after adjustment.
Retain metadata on processing dates, personnel, and reagent lots to support batch modeling.
Assess overcorrection risks when removing batch effects that may correlate with biological conditions.
Document normalization parameters and software versions for reproducibility in regulatory contexts.

Module 5: Differential Expression Analysis

Select statistical models (e.g., limma, SAM) based on sample size, design complexity, and variance stability.
Define fold-change thresholds in conjunction with p-value adjustments to prioritize biologically meaningful genes.
Apply empirical Bayes moderation of variances to improve stability in small sample studies.
Adjust for multiple testing using FDR (Benjamini-Hochberg) rather than Bonferroni to balance sensitivity and specificity.
Incorporate covariates (e.g., age, tumor stage) into linear models to isolate primary effects of interest.
Validate findings using qRT-PCR on a subset of top differentially expressed genes.
Flag genes with inconsistent probe set behavior for manual curation or exclusion.

Module 6: Functional Enrichment and Pathway Analysis

Select annotation databases (e.g., GO, KEGG, Reactome) based on organism and pathway coverage completeness.
Resolve gene identifier discrepancies between array probes and pathway databases using mapping files.
Choose between over-representation analysis (ORA) and gene set enrichment analysis (GSEA) based on hypothesis structure.
Adjust significance thresholds for redundant or correlated pathways to avoid overinterpretation.
Filter out broad GO terms (e.g., “cellular process”) that lack biological specificity.
Integrate tissue-specific expression data to prioritize relevant pathways in interpretation.
Visualize results using pathway diagrams with expression directionality and fold-change overlays.

Module 7: Data Integration and Multi-Omics Correlation

Align microarray expression data with genomic variants (e.g., SNPs) to identify expression quantitative trait loci (eQTLs).
Map probe locations to promoter regions when integrating with ChIP-seq or methylation data.
Normalize data across platforms using Z-scores or rank-based methods for combined analysis.
Apply canonical correlation analysis (CCA) to detect coordinated patterns between mRNA and protein levels.
Resolve gene symbol conflicts across datasets using authoritative sources like HGNC.
Use time-series microarray data to infer regulatory networks with dynamic Bayesian models.
Flag discordant results between microarray and RNA-seq for technical or biological investigation.

Module 8: Data Archiving and Regulatory Compliance

Format datasets according to MIAME standards for submission to public repositories (e.g., GEO, ArrayExpress).
Encrypt and store raw image files (e.g., .CEL, .TIF) for audit and reanalysis requirements.
Obtain IRB approval documentation for sharing human-derived expression data under GDPR or HIPAA.
Define data retention schedules for raw and processed files based on institutional policies.
Assign persistent identifiers (DOIs) to datasets to support citation and reproducibility.
Document all preprocessing steps in metadata using controlled vocabularies (e.g., EDAM ontology).
Restrict access to sensitive datasets using tiered permission systems in institutional databases.

Module 9: Visualization and Stakeholder Reporting

Design publication-ready heatmaps with dendrograms and annotation tracks using ComplexHeatmap in R.
Generate interactive dashboards for clinicians using Shiny to explore gene expression patterns.
Select color palettes that are perceptually uniform and accessible to colorblind users.
Summarize key findings in static summary figures for inclusion in regulatory dossiers.
Balance detail and clarity when annotating volcano plots with gene labels.
Produce dynamic reports using R Markdown or Quarto to link analysis code with visual output.
Validate figure resolution and font sizes for both digital and print publication formats.