Cancer Genome Interpreter annotates the biological and clinical relevance of tumor alterations

Abstract

While tumor genome sequencing has become widely available in clinical and research settings, the interpretation of tumor somatic variants remains an important bottleneck. Here we present the Cancer Genome Interpreter, a versatile platform that automates the interpretation of newly sequenced cancer genomes, annotating the potential of alterations detected in tumors to act as drivers and their possible effect on treatment response. The results are organized in different levels of evidence according to current knowledge, which we envision can support a broad range of oncology use cases. The resource is publicly available at http://www.cancergenomeinterpreter.org .

Fig. 1

Cancer Genome Interpreter. a Outline of the CGI workflow. With a list of genomic alterations as input, the CGI automatically recognizes the format, remaps the variants as needed, and standardizes the annotation for downstream compatibility. All analyses are cancer-specific and thus the tumor type of the sample(s) to analyze is also required. Next, the CGI identifies known driver alterations and annotates and classifies the remaining variants of unknown significance. Finally, alterations that are biomarkers of drug effects are identified. b The CGI may be run via the web at http://www.cancergenomeinterpreter.org (left panel) or through an API. The web results can be stored in a private repository (right panel) for their management. The results of the CGI are provided via interactive reports. c An example of a mutation analysis report. This contains the annotations of all mutations, which empowers the user’s review, and the labels for those known or predicted to be drivers by OncodriveMUT. d An example of a biomarker match report. This contains the putative biomarkers of drug response found in the tumor organized according to distinct levels of clinical relevance. All these web reports are interactive and configurable by the user. CNA copy number alteration

Fig. 2

Annotating mutations in cancer genes. a Catalog of Cancer Genes. Genes that drive tumorigenesis via mutations, copy number alterations, and/or translocations are annotated with their mode of action (MoA). b Catalog of Validated Oncogenic Mutations. Clinically or experimentally validated driver mutations were gathered from manually annotated resources and the cancer literature. c Proportion of validated mutations observed across the cancer genes of 6792 tumors. Cancer types: ALL acute lymphocytic leukemia, AML acute myeloid leukemia, BLCA bladder carcinoma, BRCA breast carcinoma, CLL chronic lymphocytic leukemia, CM cutaneous melanoma, COREAD colorectal adenocarcinoma, DLBC diffuse large B cell lymphoma, ESCA esophageal carcinoma, GBM glioblastoma multiforme, HC hepatocarcinoma, HNSC head and neck squamous cell carcinoma, LGG lower grade glioma, LUAD lung adenocarcinoma, LUSC lung squamous cell carcinoma, MB medulloblastoma, MM multiple myeloma, NB neuroblastoma, NSCLC non-small cell lung carcinoma, OV serous ovarian adenocarcinoma, PA pilocytic astrocytoma, PAAD pancreas adenocarcinoma, PRAD prostate adenocarcinoma, RCC renal clear cell carcinoma, SCLC small cell lung carcinoma, STAD stomach adenocarcinoma, THCA thyroid carcinoma, UCEC uterine corpus endometrioid carcinoma. d OncodriveMUT schema to estimate the oncogenic potential of the variants of unknown significance. A set of heuristic rules combines the annotations obtained for a given mutation with the knowledge about the genes (or regions thereof) in which it is observed, as retrieved from the computational analyses of sequenced cohorts

Fig. 3

Cancer Biomarkers Database. a A board of clinical and research experts gather the genomic biomarkers of drug response to be included in the Cancer Biomarkers database through periodic updates. The upper panel displays the simplified schema of the data model. The clinical/research community is encouraged to provide feedback to edit an existing entry or add a novel one by using the comment feature available in the web service. Any suggestion is subsequently evaluated by the scientific team and incorporated as appropriate. A semi-automatic pipeline annotates any novel entry to ensure the consistency of the attributes, including variant re-mapping from protein to genomic coordinates when necessary. The lower panel displays some of the 1574 biomarkers that have been collected in the current version of the database, and the pie charts summarize the content. CNA copy number alteration. b CGI analyses detect putative driver mutations in individual tumors that are rarely observed in the corresponding cancer type. When these variants are known targets of anti-cancer therapies, they may constitute tumor type re-purposing opportunities. The graph summarizes some of these potential opportunities detected by the CGI on 6792 pan-cancer tumors with exome-sequencing data, which are currently unexplored. The barplots display the overall number of tumor samples (separated by cancer type) in which they were observed. The acronym of the cancer type in which the genomic event is demonstrated to confer sensitivity to the drug is shown in parentheses following the name of the drug, and the clinical evidence of that association is represented through color circles (note that the clinical guidelines/recommendations label refers to FDA-approved or international guidelines). Targeted drugs and chemotherapies are shown separately. Cancer acronyms that are not included in the Fig. 2 legend: RA renal angiomyolipoma, BCC basal cell carcinoma, GCA giant cell astrocytoma, G glioma, MCL mantle cell lymphoma, MRT malignant rhabdoid tumor, R renal, CH chollangiocarcinoma. c Therapeutic landscape of 6792 tumors with exome-sequencing data. Fraction of tumors with genomic alterations that are biomarkers of drug response in each cancer type. Colors in the bars denote the clinical evidence supporting the effect of biomarkers in that disease (see evidence colors in b). Note that the event with evidence closest to the clinical evidence is given priority when several biomarkers of drug response co-occur in the same tumor sample. The lower part of the graph indicates the total number of samples per cancer type, detailing the number of samples in which mutation, CNA, and/or fusion data were analyzed. Cancer acronyms as in the Fig. 2 caption. d Same as c for a cohort of 17,462 tumors sequenced by targeted panels and gathered by the GENIE project. Tumors were grouped according to the most specific disease subtype available in the patient information. Cancer acronyms that are not included in the Fig. 2 legend are detailed in Additional file 2: Supplementary content