Background
Personalized cancer vaccines are emerging as one of the most promising approaches to immunotherapy of advanced cancers. However, only a small proportion of the neoepitopes generated by somatic DNA mutations in cancer cells lead to tumor rejection. Since it is impractical to experimentally assess all candidate neoepitopes prior to vaccination, developing accurate methods for predicting tumor-rejection mediating neoepitopes (TRMNs) is critical for enabling routine clinical use of cancer vaccines.
Results
In this paper we introduce Positive-unlabeled Learning using AuTOml (PLATO), a general semi-supervised approach to improving accuracy of model-based classifiers. PLATO generates a set of high confidence positive calls by applying a stringent filter to model-based predictions, then rescores remaining candidates by using positive-unlabeled learning. To achieve robust performance on clinical samples with large patient-to-patient variation, PLATO further integrates AutoML hyper-parameter tuning, classification threshold selection based on spies, and support for bootstrapping.
Conclusions
Experimental results on real datasets demonstrate that PLATO has improved performance compared to model-based approaches for two key steps in TRMN prediction, namely somatic variant calling from exome sequencing data and peptide identification from MS/MS data.
Using tandem mass spectroscopy (MS), we identified 3646 unique sequences among peptides eluted from purified Kd and Dd MHC I molecules of the BALB/c fibrosarcoma Meth A. These peptides were cross-referenced with the output of neoepitopes predicted for this tumor by our prediction pipeline CCCP (Cross Consensus Calling Platform). Eleven (11) of the eluted peptides were identified as neoepitopes and eight neoepitopes (of 11) were confirmed by targeted MS.
Each neoepitope was used to immunize BALB/c mice (twice, one week apart, using precise neoepitopes along with bone marrow-derived dendritic cells); mice were challenged with Meth A cells one week after the last immunization, and tumor growth was monitored in individual mice. In parallel, immunized mice were tested for CD8+ T cells to the neoepitopes using tetramer staining and interferon g secretion by CD8 cells.
Four of the eight neoepitopes elicited rejection of Meth A fibrosarcoma; two of the four neoepitopes elicited highly potent tumor rejection, while the other two elicited statistically significant but weaker tumor rejection. Of the two strong neoepitopes, only one elicited a measurable CD8 response. Both weak neoepitopes elicited measurable CD8 responses. Of the four neoepitopes that did not elicit tumor rejection, only one elicited a measurable CD8 response; this CD8 response was the strongest of all CD8 responses detected.
These observations indicate that MS-defined neoepitopes can be a rich source of neoepitopes that can mediate tumor rejection. Further, they highlight the fact that CD8 responses are not a good predictive surrogates for tumor rejection.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.