Notebook Posts Index

This index organizes notebook posts from the project analysis notebook that document the biomarker discovery journey. Posts are grouped by theme and linked chronologically within each theme.

All notebook posts are available at: https://resilience-biomarkers-for-aquaculture.github.io/[post-slug]/

Pipeline Development & Infrastructure

2024-09-10: RNAseq Workflow with Reference Genome (ES)

Timeline: September 2024
Summary: Initial RNAseq workflow setup using reference genome for C. gigas Dataset #1. Established baseline pipeline approach.

2024-09-11: RNAseq Workflow with De Novo Transcriptome

Timeline: September 2024
Summary: Alternative approach using de novo transcriptome assembly for species without well-annotated reference genomes.

2024-09-25: Set Up Nextflow on UW Klone and Run FetchNGS

Timeline: September 2024
Summary: Infrastructure setup for running nf-core pipelines on UW Klone HPC cluster. FetchNGS workflow for downloading public data.

2024-11-23: AWS Batch Compute Environment on Seqera (SY)

Timeline: November 2024
Summary: Cloud-based compute infrastructure setup using Seqera platform and AWS Batch for scalable RNA-seq processing.

2024-12-03: Repoint nf-core projectDir to srlab Directory (SW)

Timeline: December 2024
Summary: Infrastructure configuration adjustments for consistent nf-core pipeline execution.

2025-01-06: Creating a Nextflow Pipeline for Gene Count Analysis (SY)

Timeline: January 2025
Summary: Exploration of custom Nextflow pipeline development for gene count-based analyses beyond standard nf-core workflows.

Differential Abundance Analysis

2024-12-11: Exploring nf-core Differential Abundance Workflow (ES)

Timeline: December 2024
Related: Timeline - Dec 4, 2024
Summary: Initial exploration of nf-core/differentialabundance pipeline. Encountered compute resource limits when attempting to process 4 combined datasets.

Key lesson: Integration of heterogeneous datasets requires careful consideration of batch effects and nuances.

2025-07-01: Run nf-core Differentialabundance Pipeline on C. virg Data (SW)

Timeline: July 2025
Related: Issue #34, Timeline - July 2025
Summary: Per-dataset differential abundance analysis. Combined Study 4 (injected group) + Study 5 to increase sample size.

Key learning: Understanding normalization timing in the pipeline:

  • PCAs generated before differential abundance analysis
  • Normalization (VST, rlog) happens during DESeq2
  • Batch correction applied post-normalization

Observation: Started seeing evidence of innate trait in batch-corrected PCAs on top 500 most variable genes.

2025-08-05: Creating a GMT File for Use with GSEA (SY)

Timeline: August 2025
Related: Issue #41
Summary: Integrating Gene Set Enrichment Analysis (GSEA) with differential abundance pipeline. Created GMT file with gene descriptions for pathway-level interpretation.

2025-08-26: Step-wise Differential Abundance with Cvirg Dataset 1 (SW)

Timeline: August 2025
Related: Issue #41, Stepwise Pipeline
Summary: Implementation of two-step filtering approach:

  1. Step 1: Controls vs. treated (identify stress-responsive genes)
  2. Step 2: Resistant vs. sensitive (from Step 1 genes)

Results: Only 1 significant gene identified in Step 2

Problem: DESeq2’s VST couldn’t work properly with highly filtered gene set (~50 genes)

Key insight: Raised question about whether stepwise filtering removes innate biomarkers present in control samples.

Data Processing & Quality Control

2024-11-04: Run RNAseq on Perkinsus Datasets (SW)

Timeline: November 2024
Summary: Processing and quality control of Perkinsus marinus challenge datasets.

2024-12-13: Compare Genes After CGI ID Conversion Attempt 1 (SW)

Timeline: December 2024
Summary: Gene identifier harmonization across datasets with different annotation versions. Critical for cross-dataset comparisons.

2024-12-23: PCA for Gene Count Comparison Between NCBI Annotation Releases (SY)

Timeline: December 2024
Summary: Assessed impact of different C. gigas annotation releases on gene count matrices and downstream analysis using PCA.

2025-04-15: Reprocess TAG-seq Data with Different FastP Parameters (SW)

Timeline: April 2025
Related: Issue #26, Issue #28, Timeline - April 2025
Summary: Critical technical discovery: Johnson dataset used TAG-seq (not standard RNA-seq). Initial analysis with standard RNA-seq parameters was inappropriate.

Issues:

  • GC bias detected
  • Wrong adapter trimming parameters
  • Inappropriate quality filtering

Solution: Reran with TAG-seq-specific FastP parameters, improving data quality.

Gene Classifier & Biomarker Development

2025-09-11: Two-Script Pipeline for Gene-Expression Classifier (SY)

Timeline: August-September 2025
Status: ✅ Core methodology for 6-gene classifier
Related: Issue #44, Two-Step Classifier Pipeline
Summary: Landmark analysis - Developed validated two-step classifier approach combining datasets 1 & 5.

Methodology:

  • Step 1 (R script): Meta-analysis ranking genes by reproducibility, effect consistency, and heterogeneity across studies
  • Step 2 (Python script): Logistic regression with LASSO regularization for minimal gene panel; LOSO validation

Result: 6-gene classifier with strong separation between tolerant/sensitive phenotypes

Key innovation: Explicitly separates feature discovery from classifier construction to reduce overfitting across batches

Code: Two-step pipeline repository

ChatGPT session: Development discussion

2025-10-01: Series of DESeq2 Runs Indicates Innate DEGs for Tolerance (SY)

Timeline: September 2025
Related: Issue #53, Validation & Pitfalls
Summary: Critical biological insight - Discovered that differential expression between tolerant/sensitive samples is relatively independent of treatment status.

Finding: Biomarkers appear to be innate (constitutively different between families) rather than reactive (induced by stress)

Implication: Stepwise filtering approach (control vs. treated first) would remove these innate biomarkers

Approach:

  • Used Day 7 samples from Studies 1 & 5 (most compatible designs)
  • Ran contrasts on: Treatment, Treated-only samples, Control-only samples, Combined samples
  • Treatment contrast: only 1 DEG
  • Treated-only contrast: only 1 DEG
  • Control and treated samples showed similar tolerant vs. sensitive DEGs

Conclusion: Tolerance/sensitivity is an innate differential expression profile, common to control and treated samples

2025-10-01: Exploring Six-Gene Biomarker Across Studies (SY)

Timeline: September 2025
Related: Issue #49
Summary: Visualization and validation of the 6-gene classifier panel across multiple studies. Assesses how well genes distinguish phenotypes in independent datasets.

2026-01-05: Common Genes per LOSO Fold (SY)

Timeline: January 2026
Summary: Analysis of gene consistency across Leave-One-Study-Out (LOSO) cross-validation folds. Identifies genes that are robustly selected regardless of which study is held out.

Exploratory & Supporting Analyses

2024-12-31: Using ChatGPT to Explore Thermal Resilience Based on Gene Counts (SY)

Timeline: December 2024
Summary: Exploratory use of AI tools (ChatGPT) for hypothesis generation about thermal tolerance based on gene expression patterns. Documents AI-assisted analysis workflow.

Methylation Sequencing

2024-11-27: Run Methylseq on Klone (SW)

Timeline: November 2024
Summary: Epigenetic analysis using nf-core/methylseq pipeline on C. virginica whole-genome bisulfite sequencing (WGBS) data.

2025-04-07: Reproduce Oyster WGBS Analysis with nf-core Methylseq (SW)

Timeline: April 2025
Summary: Validation of methylseq pipeline results by comparing with published oyster WGBS analysis.

2025-05-02: Run C. virginica WGBS Data Through nf-core Methylseq (SW)

Timeline: May 2025
Summary: Processing additional C. virginica WGBS datasets for methylation biomarker discovery.

2025-08-06: Run Methylseq on M. chilensis WGBS Data (SW)

Timeline: August 2025
Summary: Expansion to Chilean blue mussel (Mytilus chilensis) methylation data to assess cross-species epigenetic patterns.

Thematic Groups Summary

🔧 Pipeline Development (6 posts)

Posts focused on infrastructure, workflow setup, and computational resource configuration.

📊 Differential Abundance (4 posts)

Posts documenting differential expression analysis attempts, including integrated, per-dataset, and stepwise approaches.

🎯 Classifier & Biomarkers (4 posts)

Posts detailing the development and validation of the 6-gene classifier, including innate vs. reactive characterization.

🔬 Data QC & Processing (4 posts)

Posts addressing data quality, annotation harmonization, and technical parameter optimization (especially TAG-seq).

🧬 Methylation (4 posts)

Posts exploring epigenetic biomarkers using WGBS data.

Chronological Navigation

2024 Posts: September November December
2025 Posts: January April May July-October

All Posts: Archive by Date

Main Guide:

Other Sources: