Cancer Cell Line Authentication (CCLA)

Cancer Cell Line Authentication (CCLA) authenticates human cancer cell lines by comparing gene expression profiles to curated CCL-specific signatures using machine learning to detect misidentification and contamination.

Key Features:

• CCL Coverage: Authenticates 1,291 human cancer cell lines derived from 28 tissues.
• Input data: Accepts gene expression profiles and NCBI GEO accession numbers.
• Gene Signatures: Uses curated cancer cell line (CCL)-specific gene signatures.
• Similarity assessment: Employs advanced machine learning techniques to measure overall similarities and distances between a query sample and reference CCLs.
• Validation datasets: Validated on microarray and RNA-Seq datasets comprising 719 samples representing 461 distinct CCLs.
• Validation performance (microarray): Achieved top-1 accuracy 96.58% and top-3 accuracy 100% on microarray validation data.
• Validation performance (RNA-Seq): Achieved top-1 accuracy 92.15% and top-3 accuracy 95.11% on RNA-Seq validation data.
• Novelty: First reported approach to authenticate cancer cell lines using gene expression data.

Scientific Applications:

• Authentication and contamination detection: Detects misidentification and contamination in cancer cell lines using gene expression signatures.
• Quality control for experiments: Validates CCL identity to improve reliability and reproducibility of cell line–based cancer research.

Methodology:

Curates CCL-specific gene signatures and applies machine learning to compute similarities and distances between query gene expression profiles and reference CCLs; validation used microarray and RNA-Seq data.