Module 4

Transcriptomics & Gene Expression

RNA-seq replaced microarrays in the early 2010s and now dominates gene-expression profiling. This module covers the pipeline: alignment (STAR, HISAT2), quantification (Salmon, kallisto pseudoalignment), differential expression (DESeq2, edgeR), and single-cell extensions (scRNA-seq, pseudotime).

1. RNA-seq Pipeline

Standard workflow: (1) QC reads with FastQC; (2) trim adaptors (fastp, Trimmomatic); (3) align to genome (STAR, HISAT2) or pseudoalign to transcriptome (Salmon, kallisto); (4) quantify transcript abundance (TPM); (5) test differential expression (DESeq2, edgeR) with appropriate sample-wise dispersion models; (6) pathway analysis (GSEA, Enrichr).

2. DESeq2 & Negative Binomial

Counts of RNA-seq reads per gene follow approximately a negative-binomial distribution (Poisson with overdispersion). DESeq2 (Love 2014) fits gene-wise dispersions, shrinks them empirically to a gene-wise prior, and tests the Wald statistic against the shrunken estimate. The shrinkage stabilises low-count gene estimates and is critical for small-sample studies.

Simulation: Volcano Plot

Python

script.py37 lines

import numpy as np, matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt

# Volcano plot from simulated DE results
rng = np.random.default_rng(0)
n = 2000
logfc = rng.normal(0, 1.2, n)
# Add some true DE genes
true_up = rng.choice(n, 60, replace=False)
true_down = rng.choice(n, 60, replace=False)
logfc[true_up] += rng.uniform(1, 4, 60)
logfc[true_down] -= rng.uniform(1, 4, 60)
pvals = 10 ** (-np.abs(logfc)*1.4 - rng.exponential(1.5, n))
pvals = np.clip(pvals, 1e-15, 1)
neg_log_p = -np.log10(pvals)

fig, ax = plt.subplots(figsize=(10, 7), facecolor='#0a0a1a')
ax.set_facecolor('#111827'); ax.tick_params(colors='#cbd5e1')
for s in ax.spines.values(): s.set_color('#334155')
# Color by significance
colors = np.where((np.abs(logfc) > 1) & (neg_log_p > 2),
                  np.where(logfc > 0, '#2dd4bf', '#f87171'),
                  '#475569')
ax.scatter(logfc, neg_log_p, c=colors, s=18, alpha=0.7)
ax.axvline(1, color='#fbbf24', ls='--', alpha=0.5)
ax.axvline(-1, color='#fbbf24', ls='--', alpha=0.5)
ax.axhline(2, color='#fbbf24', ls='--', alpha=0.5)
ax.set_xlabel('log2 fold-change', color='#cbd5e1')
ax.set_ylabel('-log10(p)', color='#cbd5e1')
ax.set_title('DESeq2-style volcano plot',
             color='#5eead4', fontweight='bold')
ax.grid(True, color='#334155', alpha=0.3)
plt.tight_layout()
plt.savefig('output.png', dpi=120, bbox_inches='tight', facecolor='#0a0a1a')
print(f'Teal up-regulated: {np.sum((logfc>1) & (neg_log_p>2))}')
print(f'Red down-regulated: {np.sum((logfc<-1) & (neg_log_p>2))}')

Click Run to execute the Python code

Code will be executed with Python 3 on the server

3. Single-Cell RNA-seq

scRNA-seq (10x Chromium, Smart-seq3, Drop-seq, SPLiT-seq) profiles thousands of individual cells. Tools: Seurat (R) and Scanpy (Python) for QC, normalisation, dimensional reduction (PCA → UMAP), clustering (Louvain/Leiden), cell-type annotation. Trajectory inference (Monocle3, PAGA, scVelo) orders cells along developmental pseudotime axes.

Key References

• Love, M. I., Huber, W. & Anders, S. (2014). “Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2.” Genome Biol., 15, 550.

• Bray, N. L. et al. (2016). “Near-optimal probabilistic RNA-seq quantification.” Nat. Biotechnol., 34, 525–527.

• Patro, R. et al. (2017). “Salmon provides fast and bias-aware quantification of transcript expression.” Nat. Methods, 14, 417–419.

• Stuart, T. et al. (2019). “Comprehensive integration of single-cell data.” Cell, 177, 1888–1902.

Share:X Reddit LinkedIn

← Module 3 Module 5: AlphaFold →