Bulk RNA-seq differential expression report

ProjectDEMO-RNA-01

Data sourceGEO GSE147507

Analysis date2026-10-01

Issued byHST GENOMICS bioinformatics

Project overview

This report is a sample deliverable built on public data to show how HST GENOMICS delivers a bulk RNA-seq differential expression project. Data: GEO series GSE147507 (Blanco-Melo et al., 2020), Series 1: primary human bronchial epithelial cells (NHBE), SARS-CoV-2 infected versus mock, three replicates each, raw count matrix as deposited.

samples

14,159

genes retained

111

DE genes (padj<0.05, |log2FC|>1)

95 / 16

up / down

Sample	Group	Assigned reads
Series1_NHBE_Mock_1	mock	11.3 M
Series1_NHBE_Mock_2	mock	10.3 M
Series1_NHBE_Mock_3	mock	16.6 M
Series1_NHBE_SARS-CoV-2_1	infected	10.2 M
Series1_NHBE_SARS-CoV-2_2	infected	9.9 M
Series1_NHBE_SARS-CoV-2_3	infected	29.3 M

Table 1. Samples and groups

Methods

Input: gene-level raw counts as deposited by the authors (STAR alignment, featureCounts); reads were not re-aligned for this demo.
Filtering: genes with fewer than 10 counts across the six samples removed; 14,159 of 21,797 genes retained, of which 13,061 passed DESeq2 independent filtering and received adjusted P values.
Differential expression: PyDESeq2 0.5.4 (Python implementation of the DESeq2 method), design ~condition, contrast infected vs mock, Wald test, Benjamini–Hochberg adjustment; significance padj < 0.05 and |log2FC| > 1.
QC: library sizes, PCA on the 500 most variable genes (log2 CPM), Spearman correlation between samples.
Enrichment: over-representation analysis of up-regulated genes against MSigDB Hallmark gene sets (Enrichr via gseapy 1.3.1).

Quality control

Figure 2. PCA. PC1 explains 62.7% of variance and separates infected from mock samples.

Figure 3. Spearman correlation between samples.

Results

At padj < 0.05 and |log2FC| > 1, 111 genes are differentially expressed: 95 up and 16 down in infected cells.

Figure 4. Volcano plot; the 15 genes with the smallest padj are labelled.

Gene	baseMean	log2FC	padj
CCL20	412.1	3.14	1.87e-77
SAA2	575.8	2.42	5.82e-76
SAA1	3316.1	2.22	1.54e-52
IL36G	271.3	2.73	1.65e-51
SPRR2D	365.2	2.98	5.60e-49
S100A8	1707.1	1.87	1.07e-48
TNFAIP3	2288.8	1.61	2.09e-47
INHBA	1216.9	1.82	5.53e-47
ICAM1	1884.7	1.86	2.81e-41
KRT6B	2803.4	1.55	3.71e-41
CFB	789.3	1.85	3.71e-41
C3	6336.0	1.51	3.85e-41
TNIP1	4659.3	1.26	2.71e-40
CXCL1	1801.7	1.42	3.96e-38
SOD2	2001.4	1.52	2.65e-34
MX1	427.7	2.51	3.19e-34
ZC3H12A	870.7	1.67	1.37e-31
S100A9	12313.2	1.11	3.89e-31
CXCL5	104.7	3.48	1.56e-29
C15orf48	897.0	1.24	2.54e-28

Table 2. Twenty DE genes with the smallest padj (full table: tables/DE_results_all_genes.csv).

Figure 5. Expression heatmap of the top 30 DE genes (row z-scores).

Comparison with the original study: only 1 of the top 20 up-regulated genes (MX1) is a classical interferon-stimulated gene; the list is dominated by chemokines and inflammatory genes (CCL20, CXCL1, CXCL5, IL36G), Hallmark interferon-gamma and interferon-alpha response sets are nevertheless significantly enriched among up-regulated genes (Table 3): interferon-stimulated genes are induced, but less prominently than chemokines. This agrees with the original study (Blanco-Melo et al., 2020), which reported a comparatively muted interferon response with strong chemokine induction.

Figure 6. MSigDB Hallmark enrichment of up-regulated genes.

Term	Overlap	Adjusted P
TNF-alpha Signaling via NF-kB	27/200	9.77e-31
Interferon Gamma Response	20/200	4.74e-20
Inflammatory Response	17/200	6.07e-16
IL-6/JAK/STAT3 Signaling	12/87	5.97e-14
Interferon Alpha Response	11/97	6.10e-12
Complement	13/200	4.60e-11
KRAS Signaling Up	13/200	4.60e-11
Allograft Rejection	10/200	1.35e-07
Coagulation	7/138	1.33e-05
Apoptosis	7/161	3.28e-05
Epithelial Mesenchymal Transition	7/200	1.20e-04
IL-2/STAT5 Signaling	5/199	5.73e-03

Table 3. Top 12 enriched terms.

Deliverables

tables/DE_results_all_genes.csv: DESeq2 results for all tested genes
tables/DE_significant_padj0.05_log2FC1.csv: significant genes
tables/enrichment_hallmark_upregulated.csv: enrichment results
tables/counts_raw_selected_samples.csv, tables/sample_sheet.csv: input counts and sample sheet
figures/*.png: publication figures (PDF/SVG on request)
run_bulk.py, report_lib.py, requirements.txt, README.txt: complete re-runnable code, parameters and instructions

Methods paragraph (for the manuscript)

Gene-level raw counts were obtained from GEO (GSE147507). Genes with fewer than 10 counts across the six samples were removed. Differential expression between infected and mock samples was tested with PyDESeq2 0.5.4 (design ~condition, Wald test) with Benjamini–Hochberg correction; genes with adjusted P < 0.05 and |log2 fold change| > 1 were called differentially expressed. Up-regulated genes were tested for over-representation in MSigDB Hallmark gene sets with Enrichr.