Data Sources

BioQuery integrates six major cancer genomics databases, all accessible through natural language queries.

Overview

Source	Description	Samples	Data Types
TCGA	The Cancer Genome Atlas	11,000+	Expression, Mutations, CNV, Survival
TARGET	Pediatric cancers	6,000+	Expression, Mutations, Clinical
GTEx	Normal tissue reference	17,000+	Expression
CCLE	Cancer cell lines	~1,000	Expression, Mutations
CPTAC	Proteomics	~1,500	Protein abundance
GENIE	Real-world clinical	40,000+	Clinical data

All data is accessed via ISB-CGC BigQuery for fast, reproducible queries.

TCGA (The Cancer Genome Atlas)

The primary data source for adult cancer genomics.

Available Cancer Types (33 types)

Code	Cancer Type	Samples
BRCA	Breast invasive carcinoma	~1,100
LUAD	Lung adenocarcinoma	~600
LUSC	Lung squamous cell carcinoma	~500
KIRC	Kidney renal clear cell carcinoma	~530
KIRP	Kidney renal papillary cell carcinoma	~290
GBM	Glioblastoma multiforme	~170
OV	Ovarian serous cystadenocarcinoma	~300
COAD	Colon adenocarcinoma	~450
STAD	Stomach adenocarcinoma	~400
PRAD	Prostate adenocarcinoma	~500
LIHC	Liver hepatocellular carcinoma	~370
PAAD	Pancreatic adenocarcinoma	~180
SKCM	Skin cutaneous melanoma	~470
HNSC	Head and neck squamous cell carcinoma	~520
BLCA	Bladder urothelial carcinoma	~400
THCA	Thyroid carcinoma	~500
UCEC	Uterine corpus endometrial carcinoma	~550

View all 33 cancer types

Data Types

RNA-seq expression: TPM-normalized, hg38 aligned
Somatic mutations: MC3 mutation calls
Clinical/survival: Overall survival, progression-free survival

Example Queries


Is EGFR expression higher in LUAD vs LUSC?
What's the TP53 mutation rate in breast cancer?
Does high DDR1 predict worse survival in kidney cancer?

TARGET (Therapeutically Applicable Research to Generate Effective Treatments)

Pediatric cancer genomics data.

Available Cancer Types

Code	Cancer Type	Samples
ALL	Acute lymphoblastic leukemia	~1,500
AML	Acute myeloid leukemia	~200
NBL	Neuroblastoma	~150
OS	Osteosarcoma	~100
WT	Wilms tumor	~120
RT	Rhabdoid tumor	~50
CCSK	Clear cell sarcoma of kidney	~20

Common Synonyms

“childhood leukemia” or “pediatric leukemia” → ALL
“neuroblastoma” → NBL
“Wilms tumor” or “nephroblastoma” → WT
“osteosarcoma” or “bone cancer in children” → OS

Example Queries


What is MYC expression in neuroblastoma?
Is MYCN amplified in neuroblastoma?
Compare TP53 in pediatric AML vs ALL

TARGET data uses the same RNA-seq pipeline as TCGA, so expression values are comparable.

GTEx (Genotype-Tissue Expression)

Normal tissue expression reference for tumor vs normal comparisons.

Available Tissues (54 tissues)

Common tissues used for comparison:

Tissue	TCGA Comparison
Breast - Mammary Tissue	BRCA
Lung	LUAD, LUSC
Kidney - Cortex	KIRC, KIRP
Liver	LIHC
Colon - Transverse	COAD
Prostate	PRAD
Thyroid	THCA
Brain - Cortex	GBM, LGG

Example Queries


Is MYC upregulated in breast cancer compared to normal?
Compare DDR1 expression in lung cancer vs normal tissue
Is TP53 higher in tumors vs normal liver?

GTEx samples are from healthy donors, while TCGA “matched normal” samples come from tissue adjacent to tumors. Results may differ.

CCLE (Cancer Cell Line Encyclopedia)

Gene expression and mutation data for ~1,000 cancer cell lines.

Available Primary Sites

Site	Cell Lines
Lung	173
Haematopoietic/lymphoid	167
Breast	62
Large intestine	58
Skin	54
Central nervous system	46
Ovary	46
Pancreas	41
Stomach	37

Data Types

Expression: RMA-normalized microarray (Affymetrix)
Mutations: Whole-exome sequencing

Example Queries


What is DDR1 expression in lung cancer cell lines?
Which cell line sites have highest EGFR expression?
Is BRAF mutated in melanoma cell lines?

CCLE expression uses RMA normalization, not TPM. Values are not directly comparable to TCGA RNA-seq data.

CPTAC (Clinical Proteomic Tumor Analysis Consortium)

Mass spectrometry-based proteomics data.

Available Cancer Types

Code	Cancer Type	Samples
CCRCC	Clear cell renal cell carcinoma	~110
GBM	Glioblastoma	~100
HNSCC	Head and neck squamous cell carcinoma	~110
LSCC	Lung squamous cell carcinoma	~110
LUAD	Lung adenocarcinoma	~110
PDA	Pancreatic ductal adenocarcinoma	~140
UCEC	Uterine corpus endometrial carcinoma	~100
BRCA	Breast cancer	~120
COAD	Colon adenocarcinoma	~110
OV	Ovarian cancer	~80

Data Types

Protein abundance: Log2 ratio vs pooled reference
Phosphoproteomics: Phosphosite quantification (available for most cancer types)

Example Queries


What is TP53 protein abundance in glioblastoma?
Compare DDR1 protein levels across CPTAC cancers
Does DDR1 mRNA correlate with protein in breast cancer?

Protein abundance is reported as log2 ratios relative to a pooled reference, not absolute concentrations.

GENIE (Genomics Evidence Neoplasia Information Exchange)

Real-world clinical genomics data from multiple cancer centers.

Overview

Patients: ~40,000+
Institutions: DFCI, MSK, VICC, JHU, and others
Data: Clinical characteristics, demographics

Available Cancer Sites

Site	Patients
Lung	~7,000
Breast	~5,000
Colon	~4,500
CNS	~2,700
Skin	~2,000
Ovary	~1,800
Pancreas	~1,600
Prostate	~1,500

Example Queries


What is the age distribution in lung cancer patients?
Clinical summary for breast cancer patients

GENIE currently provides clinical/demographic data. Genomic data (mutations, expression) coming soon.

Data Access

All BioQuery data is accessed via the ISB-CGC BigQuery public datasets. This ensures:

Speed: Sub-second query execution on terabytes of data
Reproducibility: Exact SQL queries provided for every analysis
Transparency: All tables and versions documented

BigQuery Tables Used


# TCGA
isb-cgc-bq.TCGA.RNAseq_hg38_gdc_current
isb-cgc-bq.TCGA.masked_somatic_mutation_hg38_gdc_current
isb-cgc-bq.TCGA.clinical_gdc_current

# TARGET
isb-cgc-bq.TARGET.RNAseq_hg38_gdc_current
isb-cgc-bq.TARGET.masked_somatic_mutation_hg38_gdc_current

# GTEx
isb-cgc.GTEx_v7.gene_median_tpm

# CCLE
isb-cgc-bq.CCLE.RMA_expression_hg19_current
isb-cgc-bq.CCLE.somatic_mutation_hg19_current

# CPTAC
isb-cgc-bq.CPTAC.quant_proteome_*

# GENIE
isb-cgc-bq.GENIE.clinical_gdc_current