Abstract 5820
Background
DNA sequencing to identify variants is becoming increasingly valuable in clinical settings; including matching patients to approved targeted therapies, immunotherapies, and/or clinical trials. However, accurate calling of genetic variants from sequencing still remains challenging. With little corroboration between the different tools available, patients are at risk of being treated with therapies that are unsuitable for their cancer.
Methods
Here we present a novel machine learning based method for the accurate identification of somatic variants in cancer patient tumour samples, with a neural network architecture from encoded raw sequencing read information of tumour/normal sample pairings into an image, enabling it to classify whether a variant is germline, somatic, or sequencing error. The model was trained and tested on in-silico spike-in data using bam-surgeon, and then validated on a multi-cancer and multi-center dataset and benchmarked against industry standard variant callers.
Results
The approach, called somaticNET, outperforms existing industry standard tools in sensitivity and specificity, achieving an AUROC of ∼1.00 on the bam-surgeon dataset and an AUROC of ∼0.99 on the multi-cancer multicenter dataset. The model also works faster than other variant callers, in minutes compared to hours.
Conclusions
Using the power of machine learning for accurate somatic variant calling can improve patient matching to approved therapies and clinical trials, thus ensuring patients are given the right therapy at the right time to treat their cancer.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
The authors.
Funding
Cambridge Cancer Genomics.
Disclosure
G. Dubourg-Felonneau: Shareholder / Stockholder / Stock options, Full / Part-time employment: Cambridge Cancer Genomics. D. Rebergen: Shareholder / Stockholder / Stock options, Full / Part-time employment: Cambridge Cancer Genomics. C. Parsons: Shareholder / Stockholder / Stock options, Full / Part-time employment: Cambridge Cancer Genomics. H. Thompson: Shareholder / Stockholder / Stock options, Full / Part-time employment: Cambridge Cancer Genomics. J.W. Cassidy: Leadership role, Shareholder / Stockholder / Stock options, Full / Part-time employment, Officer / Board of Directors: Cambridge Cancer Genomics. N. Patel: Leadership role, Shareholder / Stockholder / Stock options, Full / Part-time employment: Cambridge Cancer Genomics. H.W. Clifford: Leadership role, Shareholder / Stockholder / Stock options, Full / Part-time employment: Cambridge Cancer Genomics.
Resources from the same session
5517 - Molecular fingerprinting in breast cancer (BC) screening using Quantum Optics (QO) technology combined with an artificial intelligence (AI) approach applying the concept of “molecular profiles at n variables (MPnV)”: a prospective pilot study.
Presenter: Jean-Marc Nabholtz
Session: Poster Display session 3
Resources:
Abstract
2152 - Inferring the correlation between incidence rates of melanoma and the average tumor-specific epitope binding ability of HLA class I molecules in different populations
Presenter: Istvan Miklos
Session: Poster Display session 3
Resources:
Abstract
4382 - Thermal Liquid Biopsy as a Valuable Tool in Lung Cancer Screening Programs
Presenter: Alberto Rodrigo
Session: Poster Display session 3
Resources:
Abstract
2465 - Towards a screening test for cancer by circulating DNA analysis
Presenter: Rita Tanos
Session: Poster Display session 3
Resources:
Abstract
3788 - Evaluation of a successful launch of the MammaPrint and BluePrint NGS kit
Presenter: Leonie Delahaye
Session: Poster Display session 3
Resources:
Abstract
3863 - Analysis of prognostic factors on overall survival in elderly women treated for early breast cancer using data mining and machine learning
Presenter: Pierre Heudel
Session: Poster Display session 3
Resources:
Abstract
1993 - Circulating tumor cell detection in epithelial ovarian cancer using dual-component antibodies targeting EpCAM and FRα
Presenter: Na Li
Session: Poster Display session 3
Resources:
Abstract
4281 - CEUS of the breast: Is it feasible in improved performance of BI-RADS evaluation of critical breast lesions?——A multi-center prospective study in China
Presenter: Jun Luo
Session: Poster Display session 3
Resources:
Abstract
2268 - Classification of abnormal findings on ring-type dedicated breast PET for detecting breast cancer
Presenter: Shinsuke Sasada
Session: Poster Display session 3
Resources:
Abstract
4035 - Prediction of benign and malignant breast masses using digital mammograms texture features
Presenter: Cui Yanhua
Session: Poster Display session 3
Resources:
Abstract