The MTBP is a clinical decision support system that is employed in several ongoing European clinical initiatives to share and harness tumor molecular data in the context of precision cancer medicine.
The public version of the portal provides an open lightweight version of the MTBP analytical pipeline. Please note that the analyses provided in this public version are limited and do not incorporate certain variant annotation layers, ad-hoc interpretation nuances and in-house actionability flags (such as eligibility for clinical trials) that are relevant for the clinical decision-making, and thus this resource is intended for research purposes only.
The public MTBP provides reports with a general annotation of the functional and predictive relevance of the gene variants (single nucleotide variants and indels) uploaded by the user.
At this moment, the public portal supports only the interpretation of single nucleotide variants and small indels, but we are currently working on developing an interface to also incorporate the analysis of other alteration types.
The functional relevance analysis interprets whether a given gene variant can confer tumor-promoting traits. This analysis supports the identification of the individual tumor genomic drivers and enables matching gene variants to biomarkers defined by functional criteria (such as ‘activating’ mutations in a given oncogene or ‘loss of function’ alterations in a given tumor suppressor). See below for more details about this analysis.
The predictive relevance analysis matches the (functionally relevant) gene variants with cancer biomarkers (prognosis, diagnosis and drug response) reported at present. See below for more details about this analysis.
The functional relevance is an allele-centric analysis that does not take into account gene and tumor context considerations (such as cancer type and/or co-occurring alterations in the same gene or other genes). There are multiple resources that can be used to annotate the allele-centric functional relevance, but there are not well-established guidelines on how to combine them at present. Therefore, we selected distinct sources of evidence that we inferred can provide strong or very strong supporting criteria (>90% and >99% predictive value, respectively), as estimated from previous work (Tamborero et al., manuscript in preparation).
Gene variants are matched with cancer biomarkers (diagnosis, prognosis and drug response) reported at present at different levels of “genomic resolution”, as follows: nucleotide change, protein change, a categorical genomic definition (e.g. “missense variants in exon 3”) or a functional entity (e.g. “loss-of-function mutation”, following the MTBP functional interpretation), as appropriate. In addition, tumor context considerations that shape the biomarker effect, such as cancer type/subtype and level of evidence supporting the biomarker assertion, are factored in and the results are consequently reported following the European Society of Medical Oncology scale of actionability (Mateo et al, Annals of Oncology 2018).
The public version of the MTBP does not issue actionability flags that may require information that can not be deduced from the generic input employed here. Instead, the public version of the MTBP is aimed to provide a comprehensive variant annotation that supports but does not substitute a more detailed evaluation of the user. For instance, the public version does not issue genetic counseling alerts, but the results flag those variants that occur in genes of the ACMG list for secondary findings as well as the evidence supporting the potential variant’s loss-of-function effect.
The public MTBP analysis pipeline uses a combination of in-house and publicly available resources to annotate variants (see below). For the functional analysis, the sources of evidence are divided into: (a) variants with well-reported effects compatible with the variant functional relevance (or lack thereof) according to clinical and experimental studies; (b) bona fide biological assumptions; and (c) computational metrics. For the predictive analysis, specific knowledgebases aimed to curate cancer biomarkers (diagnosis, prognosis and drug response) are queried.
Among other tools and resources employed to annotate the variants, the public portal uses several knowledgebases created by international initiatives whose content is open for academic research. Data models and variant syntax are first harmonized to enable their accurate aggregation, and additional filtering may be applied (e.g. to separate weaker supporting evidence, as annotated by the own knowledgebase metadata) as appropriate. At present, the knowledgebases harmonized in the public portal are ClinVar, BRCA-Exchange, OncoKB, CIViC and CGI.
As the knowledgebases data harmonization cannot be fully automated, their content is downloaded in order to perform several manual steps that facilitate the reformatting process (except OncoKB, see below). This process is performed periodically -- for more details, please note that the version of all the resources used to annotate the variants as well as the reference and access to the original evidence assertions always appear detailed in the MTBP reports. Also, note the “News” section in the public MTBP website, where updates/news of interest for the users of the public portal are posted.
The public version of the MTBP is now accessing the OncoKB content via the OncoKB web API. Note that the programmatic access to the OncoKB knowledgebase requires a license (free for academic research). We thus encourage the users of the public MTBP to register at the OncoKB website and obtain the corresponding token, which can be then passed to the MTBP when uploading the variants to be analysed. However, OncoKB developers kindly facilitated that, in case that the user of the public MTBP does not pass that token, the MTBP reports will still display the OncoKB content with less detailed annotation.
The public version of the MTBP only supports the analysis of single nucleotide variants and small indels at the moment. Variants can be uploaded in a VCF file or by using a free text box. For the latter, HGVS nomenclature with several reference systems is accepted; please see the content of the help popup available in the ‘Analyse variants’ interface for further information.
The MTBP provides the results in a HTML report that classifies the variants in three different tables according to their functional relevance classification (putative functionally relevant, unclassified and putative functionally neutral). Please note the distinct sources of evidence supporting that functional classification as well as the actionability tiering for variants matching with reported cancer biomarkers (diagnosis, prognosis and drug response) following the ESMO scale of actionability. Also note the use of interactive elements in the HTML report to (i) open pop-up windows with further information and variant annotation details; (ii) access to external resources with the original evidence; (iii) jump to a gene-detailed view displaying the overall characteristics of the cancer mutations in the affected gene.
The public MTBP should work well with the following web browsers: Chrome Version v71, Microsoft Edge, FireFox Quantum, Opera V57. Please note that Internet Explorer & Safari are not supported, and you can experience visualization problems with them.
The MTBP is an academic clinical decision support system and a data sharing infrastructure utilized by several European clinical initiatives. The public version of the portal is a lightweight version created as an open resource for the community with research purposes only. The access to the public MTBP is provided via the website https://www.mtbp.org/. Other access possibilities (e.g. programmatic and/or local installations) and/or inclusion of additional features beyond the general framework provided by the public version may require specific considerations and/or development; please contact firstname.lastname@example.org for further information.
When using the MTB portal, please consider citing: Tamborero D, Dienstmann R et al. Nature Medicine 2020.