pirlygenes 3.15.0

Gene lists for cancer immunotherapy expression analysis

pirlygenes

Gene lists and expression data for cancer immunotherapy

Quick start

pip install pirlygenes

# Generate all plots for a sample
pirlygenes plot-expression gene_expression_salmon.tsv

# Prostate-cancer-specific analysis
pirlygenes plot-expression gene_expression_salmon.tsv --cancer-type prostate

# List all bundled datasets and cancer types
pirlygenes data

Pan-cancer expression data

Expression data for ~3,100 therapy-relevant genes across 50 normal tissues (HPA v23 consensus nTPM) and 33 TCGA cancer types (median FPKM/TPM). All data ships with the package — no downloads needed.

Cancer types

Code	Cancer type	Aliases
ACC	Adrenocortical Carcinoma	adrenocortical, adrenal
BLCA	Bladder Urothelial Carcinoma	bladder
BRCA	Breast Invasive Carcinoma	breast
CESC	Cervical Squamous Cell Carcinoma	cervical, cervix
CHOL	Cholangiocarcinoma	cholangiocarcinoma, bile_duct
COAD	Colon Adenocarcinoma	colon, colorectal
DLBC	Diffuse Large B-Cell Lymphoma	dlbcl, lymphoma
ESCA	Esophageal Carcinoma	esophageal, esophagus
GBM	Glioblastoma Multiforme	glioblastoma, gbm
HNSC	Head and Neck Squamous Cell Carcinoma	head_neck, hnscc
KICH	Kidney Chromophobe	kidney_chromophobe
KIRC	Kidney Renal Clear Cell Carcinoma	kidney, kidney_clear
KIRP	Kidney Renal Papillary Cell Carcinoma	kidney_papillary
LAML	Acute Myeloid Leukemia	leukemia, aml
LGG	Brain Lower Grade Glioma	glioma, lgg, low_grade_glioma
LIHC	Liver Hepatocellular Carcinoma	liver
LUAD	Lung Adenocarcinoma	lung_adeno
LUSC	Lung Squamous Cell Carcinoma	lung_squamous
MESO	Mesothelioma	mesothelioma
OV	Ovarian Serous Cystadenocarcinoma	ovarian, ovary
PAAD	Pancreatic Adenocarcinoma	pancreatic, pancreas
PCPG	Pheochromocytoma and Paraganglioma	pheochromocytoma, paraganglioma
PRAD	Prostate Adenocarcinoma	prostate
READ	Rectum Adenocarcinoma	rectal
SARC	Sarcoma	sarcoma
SKCM	Skin Cutaneous Melanoma	melanoma, skin
STAD	Stomach Adenocarcinoma	stomach, gastric
TGCT	Testicular Germ Cell Tumor	testicular, testis
THCA	Thyroid Carcinoma	thyroid
THYM	Thymoma	thymoma
UCEC	Uterine Corpus Endometrial Carcinoma	endometrial, uterine
UCS	Uterine Carcinosarcoma	uterine_carcinosarcoma
UVM	Uveal Melanoma	uveal_melanoma

Python API

from pirlygenes.gene_sets_cancer import (
    pan_cancer_expression,      # full expression matrix
    cancer_types,               # list of available TCGA codes
    cancer_expression,          # expression for one cancer type
    cancer_enriched_genes,      # genes enriched in one cancer type
    cancer_surfaceome_evidence, # tumor-specific surface targets (TCSA)
    surface_protein_evidence,   # all surface proteins (SURFY/CSPA)
)

# Full matrix: ~3,100 genes x 83 columns (50 tissues + 33 cancers)
df = pan_cancer_expression()

# Housekeeping-normalized, log-scaled (for heatmaps)
df = pan_cancer_expression(normalize="housekeeping", log_transform=True)

# Expression in prostate cancer (housekeeping-normalized)
df = cancer_expression("prostate")

# Genes enriched in prostate vs other cancers (>3x fold-change)
df = cancer_enriched_genes("prostate", min_fold=3.0)

# Specific genes across all cancers
df = pan_cancer_expression(genes=["EGFR", "ERBB2", "MSLN", "FOLR1"])

Surface proteins

from pirlygenes.gene_sets_cancer import (
    surface_protein_gene_names,       # 2,799 surfaceome genes
    surface_protein_gene_ids,
    cancer_surfaceome_gene_names,     # 147 tumor-specific surface targets
    cancer_surfaceome_evidence,       # with druggability + cancer types
)

# CSPA mass-spec validated only
validated = surface_protein_gene_names(validated_only=True)  # 1,410 genes

Plotting

from pirlygenes import plot_gene_expression, plot_sample_vs_cancer

# Strip plot: gene expression by category
plot_gene_expression(df_expr, save_to_filename="summary.png")

# Scatter: sample vs cancer cohort (PDF + per-category PNGs)
plot_sample_vs_cancer(df_expr, cancer_type="prostate", save_to_filename="vs_prad.pdf")

---

TCR-T

Clinical trials

Last updated: September 17th, 2024

Sources:

• Toward a comprehensive solution for treating solid tumors using T-cell receptor therapy: A review

CAR-T

Approved therapies

Last updated: September 17th, 2024

Sources:

• CAR-T: What Is Next?

Multi-specific antibodies and T-cell engagers

Clinical trials

Last updated: September 11th, 2024

Sources:

• Progresses of T-cell-engaging bispecific antibodies in treatment of solid tumors

Antibody-drug conjugates (ADCs)

Approved

Last updated: September 19th, 2024

Sources:

• Development of antibody-drug conjugates in cancer: Overview and prospects

Clinical trials

Last updated: September 11th, 2024

Sources:

• Pan-cancer analysis of antibody-drug conjugate targets and putative predictors of treatment response

Radioligand therapies (RLTs)

Current target list

Last updated: February 11th, 2026

Sources:

• Radioligand therapy in precision oncology: opportunities and challenges
• FDA approves Pluvicto for metastatic castration-resistant prostate cancer
• FDA approves Lutetium Lu 177 dotatate for gastroenteropancreatic neuroendocrine tumors
• Emerging molecular targets and agents in radioligand therapy for solid tumors
• Early evidence for anti-CD20 targeted alpha-radiation approaches in B-cell malignancies

Methodology:

• pirlygenes/data/radioligand-targets.csv is a curated target-level list (gene targets, Ensembl IDs, and target status buckets) intended to power gene-set visualization while trial-level v1.4.0 curation is in progress.

CLI plotting notes:

• Treatment plots now include a Radio category label (capitalized consistently with other treatment labels).
• Use --label-genes to force annotation of genes that should always be text-labeled, for example: --label-genes FAP,CD276.
• PNG output defaults are larger/higher resolution (--plot-height 12.0, --plot-aspect 1.4, --output-dpi 300), and can be overridden from CLI.

Cancer-testis antigens (CTAs)

Last updated: March 23rd, 2026

Quick start

from pirlygenes.gene_sets_cancer import (
    CTA_gene_names,              # recommended: filtered, reproductive-restricted CTAs
    CTA_gene_ids,                # same, as Ensembl gene IDs
    CTA_unfiltered_gene_names,   # full superset from all source databases
    CTA_unfiltered_gene_ids,     # same, as Ensembl gene IDs
    CTA_evidence,                # full DataFrame with all evidence columns
)

# Default: expressed, reproductive-restricted CTAs (~257 genes)
cta_genes = CTA_gene_names()

# Full unfiltered superset from all sources (~358 genes)
all_ctas = CTA_unfiltered_gene_names()

# Partition ALL protein-coding genes into CTA / never-expressed / non-CTA
from pirlygenes.gene_sets_cancer import CTA_partition_gene_ids
p = CTA_partition_gene_ids()   # p.cta, p.cta_never_expressed, p.non_cta

# Evidence table with per-gene HPA tissue restriction data
df = CTA_evidence()

Pipeline overview

The CTA gene set is built as an unbiased union of genes from multiple CT antigen databases and literature sources, then systematically filtered using Human Protein Atlas tissue expression data.

Step 1: Collect — union of protein-coding CT genes from multiple source databases (358 genes):

Source	Genes	Reference
CTpedia	167	Almeida et al. 2009, NAR
CTexploreR/CTdata	62 new	Loriot et al. 2025, PLOS Genetics
Protein-level CT genes (136 total, 46 overlap)	89 new	da Silva et al. 2017, Oncotarget
EWSR1-FLI1 CT gene binding sites	12	Gallegos et al. 2019, Mol Cell Biol
Meiosis, piRNA, spermatogenesis genes	28	Multiple sources (see docs)

Each gene is tracked with a source_databases column indicating which databases include it (CTpedia, CTexploreR_CT, CTexploreR_CTP, daSilva2017, daSilva2017_protein). Only protein-coding genes (Ensembl biotype) are included. Genes with outdated HGNC symbols are renamed to current symbols with old names kept as aliases.

Step 2: Annotate — each gene is scored against Human Protein Atlas v23 tissue expression:

• RNA: HPA RNA tissue consensus (rna_tissue_consensus.tsv) — normalized transcripts per million (nTPM) across 50 normal tissues
• Protein: HPA normal tissue IHC (normal_tissue.tsv) — immunohistochemistry detection levels (Not detected / Low / Medium / High) across 63 tissues with antibody reliability scores (Enhanced / Supported / Approved / Uncertain)

Step 3: Filter — protein-coding + tiered thresholds based on protein antibody confidence (278 of 358 pass):

Protein evidence	Deflated RNA threshold
Enhanced (orthogonal validation)	≥ 80%
Supported (consistent characterization)	≥ 90%
Approved (basic validation)	≥ 95%
Uncertain or no protein data	≥ 99%

Genes with protein detected in non-reproductive tissues always fail. Thymus is excluded from all restriction checks (AIRE-driven mTEC expression is expected for CTAs).

Gene set counts

Function	Description	Count
CTA_gene_names()	Recommended default. Expressed, reproductive-restricted CTAs	~257
CTA_never_expressed_gene_names()	CTAs from databases but no HPA expression (max nTPM < 2, no protein)	~21
CTA_filtered_gene_names()	All filter-passing CTAs (= expressed + never_expressed)	~278
CTA_excluded_gene_names()	CTAs that fail filter (somatic expression)	~80
CTA_unfiltered_gene_names()	Full superset from all source databases	358
CTA_evidence()	Full DataFrame with all evidence columns	358 rows
CTA_partition_gene_ids()	Partition all protein-coding genes (dataclass with .cta, .cta_never_expressed, .non_cta sets)	~20k
CTA_partition_gene_names()	Same, as gene symbols	~20k
CTA_partition_dataframes()	Same, as DataFrames with evidence columns	~20k

Evidence columns

Each gene in cancer-testis-antigens.csv carries identity and HPA-derived evidence:

Column	Description
Ensembl_Gene_ID	Ensembl gene ID (validated against release 112)
source_databases	Semicolon-separated list of source databases (CTpedia, CTexploreR_CT, CTexploreR_CTP, daSilva2017)
biotype	Ensembl gene biotype (must be protein_coding to pass filter)
Canonical_Transcript_ID	Longest protein-coding transcript (Ensembl 112)
protein_reproductive	IHC detected only in {testis, ovary, placenta} (excl. thymus), or "no data"
protein_thymus	IHC detected in thymus
protein_reliability	Best HPA antibody reliability: Enhanced / Supported / Approved / Uncertain / "no data"
protein_strict_expression	Semicolon-separated tissues with IHC detection (excl. thymus)
rna_reproductive	All tissues with ≥1 nTPM (excl. thymus) are in {testis, ovary, placenta}
rna_thymus	Thymus nTPM ≥ 1
rna_reproductive_frac	Fraction of total nTPM (excl. thymus) in core reproductive tissues
rna_deflated_reproductive_frac	(1 + Σ_repro max(0, nTPM−1)) / (1 + Σ_all max(0, nTPM−1))
rna_deflated_reproductive_and_thymus_frac	Same but thymus added to reproductive numerator
rna_80/90/95/99_pct_filter	Whether deflated reproductive fraction ≥ threshold
filtered	Final inclusion flag (see tiered thresholds above)

For full details on the curation process, evidence columns, and filter logic, see docs/cta-curation.md.

Deflated RNA metric

The deflated metric max(0, nTPM − 1) per tissue suppresses low-level basal transcription noise before computing the reproductive fraction. A +1 pseudocount on numerator and denominator prevents 0/0 for very-low-expression genes. Example: CTCFL/BORIS has raw reproductive fraction 54% (diluted by sub-1 nTPM noise across ~40 tissues) but deflated fraction 100% (only testis has ≥1 nTPM).

Class I MHC antigen presentation

Last updated: July 21st, 2018

Sources:

• Frequent HLA class I alterations in human prostate cancer: molecular mechanisms and clinical relevance:
  • LMP2/7
  • peptide transporters TAP1/2
  • chaperones calreticulin, calnexin, ERP57, and tapasin
  • IFR-1 and NLRC5
• Expression of Antigen Processing and Presenting Molecules in Brain Metastasis of Breast Cancer
  • "β2-microgloblin, transporter associated with antigen processing (TAP) 1, TAP2 and calnexin are down-regulated in brain lesions compared with unpaired breast lesions"
• NLRC5/MHC class I transactivator is a target for immune evasion in cancer
• TAPBPR: a new player in the MHC class I presentation pathway

Interferon-gamma response

Last updated: July 21st, 2018

Sources:

• Interferon Receptor Signaling Pathways Regulating PD-L1 and PD-L2 Expression
  • "JAK1/JAK2-STAT1/STAT2/STAT3-IRF1 axis primarily regulates PD-L1 expression, with IRF1 binding to its promoter"
  • "PD-L2 responded equally to interferon beta and gamma and is regulated through both IRF1 and STAT3, which bind to the PD-L2 promoter"
  • "the suppressor of cytokine signaling protein family (SOCS; mostly SOCS1 and SOCS3) are involved in negative feedback regulation of cytokines that signal mainly through JAK2 binding, thereby modulating the activity of both STAT1 and STAT3"
• Mutations Associated with Acquired Resistance to PD-1 Blockade in Melanoma
  • "resistance-associated loss-of-function mutations in the genes encoding interferon-receptor–associated Janus kinase 1 (JAK1) or Janus kinase 2 (JAK2), concurrent with deletion of the wild-type allele"
• SOCS, inflammation, and cancer
  • "Abnormal expression of SOCS1 and SOCS3 in cancer cells has been reported in human carcinoma associated with dysregulation of signals from cytokine receptors"

Recurrently mutated cancer genes

Last updated: July 21st, 2018

Cancer genes and recurrent mutations extract from Comprehensive Characterization of Cancer Driver Genes and Mutations.

Genes extracted from Table S1 into cancer-driver-genes.csv. Mutations extracted from Table S4 into cancer-driver-variants.csv.

Both datasets were annotated with Ensembl IDs using Ensembl release 92.