Search

Genomics involving tens of thousands of genes is a complex system determining phenotype. An interesting and vital issue is how to integrate highly sparse genetic genomics data with a mass of minor effects into a prediction model for improving prediction power. We find that the deep learning method can work well to extract features by transforming highly sparse dichotomous data to lower-dimensional continuous data in a non-linear way. This may provide benefits in risk prediction-associated genotype data. We developed a multi-stage strategy to extract information from highly sparse binary genotype data and applied it for cancer prognosis. Specifically, we first reduced the size of binary biomarkers via a univariable regression model to a moderate size. Then, a trainable auto-encoder was used to learn compact features from the reduced data. Next, we performed a LASSO problem process to select the optimal combination of extracted features. Lastly, we applied such feature combination to real cancer prognostic models and evaluated the raw predictive effect of the models. The results indicated that these compressed transformation features could better improve the model’s original predictive performance and might avoid an overfitting problem. This idea may be enlightening for everyone involved in cancer research, risk reduction, treatment, and patient care via integrating genomics data.

PurposeWe developed a strategy of building prognosis gene signature based on clinical treatment responsiveness to predict radiotherapy survival benefit in breast cancer patients.Methods and MaterialsAnalyzed data came from the public database. PFS was used as an indicator of clinical treatment responsiveness. WGCNA was used to identify the most relevant modules to radiotherapy response. Based on the module genes, Cox regression model was used to build survival prognosis signature to distinguish the benefit group of radiotherapy. An external validation was also performed.ResultsIn the developed dataset, MEbrown module with 534 genes was identified by WGCNA, which was most correlated to the radiotherapy response of patients. A number of 11 hub genes were selected to build the survival prognosis signature. Patients that were divided into radio-sensitivity group and radio-resistant group based on the signature risk score had varied survival benefit. In developed dataset, the 3-, 5-, and 10-year AUC of the signature were 0.814 (CI95%: 0.742–0.905), 0.781 (CI95%: 0.682–0.880), and 0.762 (CI95%: 0.626–0.897), respectively. In validation dataset, the 3- and 5-year AUC of the signature were 0.706 (CI95%: 0.523–0.889) and 0.743 (CI95%: 0.595–0.891). The signature had higher predictive power than clinical factors alone and had more clinical prognosis efficiency. Functional enrichment analysis revealed that the identified genes were mainly enriched in immune-related processes. Further immune estimated analysis showed the difference in distribution of immune micro-environment between radio-sensitivity group and radio-resistant group.ConclusionsThe 11-gene signature may reflect differences in tumor immune micro-environment that underlie the differential response to radiation therapy and could guide clinical-decision making related to radiation in breast cancer patients.