Case Studies

Featured Projects

Real-world bioinformatics solutions delivered for biotech, pharma, and academic clients. All projects follow best practices for reproducibility, documentation, and scientific rigor.

🧬 113-Gene Classifier for Lung Adenocarcinoma

Client Challenge

Roche/Genentech needed to identify patient subgroups in lung adenocarcinoma who would respond to MEK inhibitors. Over 800 patient samples across multiple clinical trials (TCGA, OAK, POPLAR) required comprehensive transcriptional subtyping.

Our Solution

Developed a 113-gene PAM (Prediction Analysis for Microarrays) classifier using consensus nonnegative matrix factorization (NMF) to identify three distinct molecular subtypes (MUC, PRO, MES) with differential drug response.

Technical Approach

Data Integration:

Harmonized >800 patient samples across TCGA, OAK (Phase III), POPLAR, and gp28363 trials
Batch effect correction and normalization
Multi-cohort validation design

Analysis Methods:

Consensus NMF for subtype discovery
PAM classifier with 113-gene signature
Cross-validation with independent cohorts (87-91% accuracy)
Survival analysis (Cox proportional hazards)
Gene set enrichment (Camera + MSigDB Hallmark)
Drug response modeling (526 compounds, 89 cell lines)

Validation:

175 cell lines, 232 PDX models, 108 GEMM tumors
Cross-cohort validation (4-fold larger validation set)
Independent clinical trial testing

Results & Impact

✅ Published in Clinical Cancer Research (2021) — High-impact journal
✅ 87-91% accuracy across independent validation cohorts
✅ Identified MES subtype as primary beneficiary of MEK inhibition
✅ Clinical utility for patient stratification in future trials
✅ Open source code available on GitHub

Technologies: R/Bioconductor, NMF, PAM, limma, survfit, ggplot2, ComplexHeatmap

📄 Read Publication

🔬 Genome-Wide CRISPR Screen Analysis Platform

Client Challenge

Identify genes involved in tumor immune evasion using a genome-wide CRISPR knockout screen with ~160,000 sgRNAs. Needed robust analysis pipeline for hit calling, validation, and mechanism characterization.

Our Solution

Comprehensive CRISPR screen analysis pipeline using crisprVerse workflow with TMM normalization on non-essential genes, limma-voom for differential abundance, and integration with ChIP-seq and RNA-seq data.

Technical Approach

Screen Analysis:

~160,000 sgRNAs processed (screenCounter)
TMM normalization on non-essential genes
Robust empirical Bayes statistics (limma-voom)
FDR correction for multiple testing
Hit prioritization based on fold-change and statistical significance

Mechanistic Validation:

RNA-seq analysis (HTSeqGenie pipeline)
ChIP-seq integration (public GEO data: GSE102616)
Pathway enrichment (MSigDB Hallmark)
TCGA survival modeling (Cox proportional hazards)
Clinical correlation analysis

Results & Impact

✅ Published in iScience (2025)
✅ Identified ZFX transcription factor as key regulator of immune evasion
✅ Biomarker discovery for anti-PD-1 response prediction
✅ Mechanistic insights into apoptosis pathway regulation
✅ Public data available (GEO: GSE276964, GSE276965)

Technologies: crisprVerse, screenCounter, limma-voom, edgeR, BWA, TCGA integration

📄 Read Publication

🧫 Single-Cell RNA-seq Pipeline for CRISPR Screens

Client Challenge

Academic and pharma clients needed reproducible, automated pipeline for processing single-cell CRISPR perturbation screens, generating ML-ready datasets with proper quality control and normalization.

Our Solution

Developed production-grade Snakemake pipeline automating the entire workflow from FASTQ to analysis-ready perturbation datasets. The pipeline implements comprehensive QC, normalization using Scanpy, and balanced class sampling for downstream machine learning applications.

Technical Approach

Pipeline Features:

Automated quality control and filtering
Scanpy-based normalization workflow
Balanced control/perturbation group sampling
Integration of multiple samples and batches
Dimensionality reduction (PCA, UMAP)
Cell type annotation support
Multiple output formats (CSV, h5ad, AnnData)
ML-ready feature matrices

Reproducibility:

Complete Docker containerization
Snakemake workflow management
Comprehensive configuration files
Automated Quarto report generation
Version-controlled codebase (GitHub)
Continuous integration testing (GitHub Actions)

Key Technologies:

Snakemake for workflow orchestration
Python/Scanpy for scRNA-seq analysis
Docker for reproducible environments
Quarto for dynamic documentation
GitHub Actions for CI/CD

Results & Impact

✅ Open source on GitHub
✅ Fully documented with comprehensive guide
✅ Production-ready for multiple client projects
✅ Scalable from hundreds to tens of thousands of cells
✅ Reproducible across HPC and cloud environments
✅ Flexible configuration for different experimental designs

Technologies: Snakemake, Python, Scanpy, Docker, Quarto, GitHub Actions

🔗 GitHub Repository 📖 Documentation

💊 NK Cell Activation Analysis for Bispecific Antibody

Client Challenge

Characterize NK cell activation mechanisms following T cell-dependent bispecific antibody (TDB) treatment using RNA-seq, flow cytometry, and multiplex cytokine analysis.

Our Solution

Integrated multi-modal analysis combining RNA-seq (Smart-Seq V4), flow cytometry, and Luminex cytokine profiling to identify IFN, TNF, and IL2/IL10 signaling axes driving NK cell activation.

Technical Approach

RNA-seq Analysis:

Smart-Seq V4 ultra-low input (2ng total RNA)
HTSeqGenie processing pipeline
edgeR (logCPM) and voom/limma differential expression
Baseline vs 24h TDB treatment comparison

Pathway Analysis:

fgsea with MSigDB collections (c2, c5, c7)
IFN, TNF, interleukin pathway enrichment
Cytokine receptor signaling analysis

Data Integration:

Flow cytometry gating and quantification
Luminex multiplex cytokine data (17-plex panel)
Correlation analysis across data types

Results & Impact

✅ Published in Cancer Immunology Research (2024)
✅ Identified IL2/IL10 axis as key NK activation mechanism
✅ Enhanced ADCC observed in vitro and in vivo
✅ Mechanistic understanding for TDB development

Technologies: HTSeqGenie, edgeR, limma-voom, fgsea, FlowJo, Luminex

📄 Read Publication

🏥 Multi-Omics Integration for Tumor Microenvironment

Client Challenge

Understand KRAS-driven lung cancer immune escape mechanisms using integrated analysis of bulk RNA-seq, single-cell RNA-seq, and whole exome sequencing (WES).

Our Solution

Multi-modal analysis combining temporal bulk RNA-seq (8w, 12w, endpoint), 10X Genomics scRNA-seq (40,000 cells, 22 clusters), and WES (COSMIC signature analysis) to map immune evasion pathways.

Technical Approach

Bulk RNA-seq:

Multiple timepoints (longitudinal analysis)
Metacore GeneGO pathway analysis
Differential expression across treatment groups

Single-Cell Analysis:

10X Genomics Chromium platform
22 clusters identified using ImmGen databrowser
PANTHER overrepresentation analysis
EGFR/ERBB signaling pathway enrichment (33.5x, FDR 1.06e-2)

WES Integration:

COSMIC signature analysis (APOBEC mutagenesis)
Mutation burden assessment
Driver mutation identification

Results & Impact

✅ Preprint on bioRxiv (2023)
✅ Identified ERBB ligand upregulation (AREG, HBEGF) as immune escape mechanism
✅ Afatinib restores immune infiltration post-αPD-1 resistance
✅ Therapeutic strategy for combination therapy

Technologies: 10X Genomics, Seurat, ImmGen, PANTHER, GATK, COSMIC

📄 Read Preprint

📦 R Package Development for Pharma Workflows

Client Challenge

Internal pharma teams needed standardized, documented R package for Fluidigm qPCR data analysis with comprehensive testing and version control.

Our Solution

Developed production-grade R package following CRAN/Bioconductor standards with comprehensive documentation (roxygen2), unit tests (testthat >80% coverage), and pkgdown website.

Technical Approach

Package Development:

devtools/usethis workflow
roxygen2 documentation for all functions
testthat unit tests (>80% coverage)
Vignettes with worked examples
S3/S4 class systems for data structures
Error handling and input validation

Documentation:

pkgdown website with GitHub Pages
Comprehensive function reference
Multiple vignettes (basic usage, advanced workflows)
NEWS.md for version tracking

Quality Assurance:

R CMD check passes with no errors/warnings
Continuous integration (GitHub Actions)
Code coverage monitoring
CRAN standards compliance

Results & Impact

✅ Internal deployment at Roche/Genentech
✅ Standardized workflows across analysis teams
✅ Reduced analysis time through automation
✅ Comprehensive documentation reducing onboarding time
✅ Maintainable codebase with unit tests

Technologies: R, devtools, testthat, roxygen2, pkgdown, GitHub Actions

💡 Interested in Similar Projects?

We can apply these same methodologies and best practices to your research:

🧬 NGS Analysis — RNA-seq, scRNA-seq, ChIP-seq, CRISPR screens, WGS/WES

📦 R Package Development — Custom packages with testing and documentation

⚙️ Pipeline Development — Snakemake/Nextflow with Docker containers

🤖 AI Integration — ML-ready datasets and predictive modeling

📊 Multi-Omics Analysis — Integrated analysis of multiple data types

📧 Discuss Your Project 🛠️ View All Services

📚 Research Output

All case studies above resulted in peer-reviewed publications demonstrating scientific rigor and impact.

📖 View Full Publication List

🚀 Ready to Start Your Project?

Let’s discuss how we can apply our expertise to your bioinformatics challenges.

📧 Contact Us 💼 LinkedIn

Email: kontakt@actn3.pl
Response Time: Within 24 hours