CIViCpy

CIViCpy provides programmatic access to the Clinical Interpretations of Variants in Cancer (CIViC) knowledgebase for extraction and high-throughput annotation of genomic variant interpretations to support precision oncology.

Key Features:

• Python Interface: Programmatic access to CIViC representing Assertions, Evidence, Variants, and Genes as dynamic Python objects.
• High-Throughput Annotation: Annotates approximately 1,200 variants per second and matches variants to CIViC evidence levels A (professional guidelines) and B (clinical trials).
• Interoperability and Analysis: Provides reusable library functionality to integrate CIViC data into third-party analytical workflows.

Scientific Applications:

• Variant annotation and clinical interpretation: Systematic matching of observed tumor variants with CIViC-curated clinical evidence to support precision oncology decision-making.
• Large-cohort evaluation: Applied to the AACR Project GENIE dataset of 59,437 sequenced tumors, yielding precise variant matches for 38.6% of cases.

Methodology:

Programmatic querying of CIViC records (Assertions, Evidence, Variants, Genes) as dynamic Python objects and high-throughput annotation (~1,200 variants/s) enabling matching to CIViC evidence levels A and B.