Increasing evidence suggests circular RNAs (circRNAs) exert critical functions in tumor progression via sponging miRNAs (microRNAs). However, the role of circRNAs in breast cancer remains unclear. Here we systematically analyzed the circular RNAs in breast cancer based on their characteristic in sponging disease-specific miRNAs and identified hsa_circ_001783 as a top ranked circRNA in our computation and verified its high expression in both breast cancer cells and cancer tissue. A higher level of hsa_circ_001783 was significantly correlated with heavier tumor burden and poorer prognosis of patients with breast cancer. Knockdown of this circRNA remarkably inhibited the proliferation and invasion of breast cancer cells. Importantly, hsa_circ_001783 promoted progression of breast cancer cells via sponging miR-200c-3p. Taken together, hsa_circ_001783 may serve as a novel prognostic and therapeutic target for breast cancer.

Image compression-based approaches for defending against the adversarial-example attacks, which threaten the safety use of deep neural networks (DNN), have been investigated recently. However, prior works mainly rely on directly tuning parameters like compression rate, to blindly reduce image features, thereby lacking guarantee on both defense efficiency (i.e. accuracy of polluted images) and classification accuracy of benign images, after applying defense methods. To overcome these limitations, we propose a JPEG-based defensive compression framework, namely "feature distillation", to effectively rectify adversarial examples without impacting classification accuracy on benign data. Our framework significantly escalates the defense efficiency with marginal accuracy reduction using a twostep method: First, we maximize malicious features filtering of adversarial input perturbations by developing defensive quantization in frequency domain of JPEG compression or decompression, guided by a semi-analytical method; Second, we suppress the distortions of benign features to restore classification accuracy through a DNN-oriented quantization refine process. Our experimental results show that proposed "feature distillation" can significantly surpass the latest input-transformation based mitigations such as Quilting and TV Minimization in three aspects, including defense efficiency (improve classification accuracy from ∼ 20% to ∼ 90% on adversarial examples), accuracy of benign images after defense (≤ 1% accuracy degradation), and processing time per image (∼ 259× Speedup). Moreover, our solution can also provide the best defense efficiency (∼ 60% accuracy) against the recent adaptive attack with least accuracy reduction (∼ 1%) on benign images when compared with other input-transformation based defense methods.

Background: Although both circular RNAs (circRNAs) and autophagy are associated with the function of breast cancer (BC), whether circRNAs regulate BC progression via autophagy remains unknown. In this study, we aim to explore the regulatory mechanisms and the clinical significance of autophagy-associated circRNAs in BC. Methods: Autophagy associated circRNAs were screened by circRNAs deep sequencing and validated by qRT-PCR in BC tissues with high-and low-autophagic level. The biological function of autophagy associated circRNAs were assessed by plate colony formation, cell viability, transwells, flow cytometry and orthotopic animal models. For mechanistic study, RNA immunoprecipitation, circRNAs pull-down, Dual luciferase report assay, Western Blot, Immunofluorescence and Immunohistochemical staining were performed. Results: An autophagy associated circRNA circCDYL was elevated by 3.2 folds in BC tissues as compared with the adjacent non-cancerous tissues, and circCDYL promoted autophagic level in BC cells via the miR-1275-ATG7/ULK1 axis; Moreover, circCDYL enhanced the malignant progression of BC cells in vitro and in vivo. Clinically, increased circCDYL in the tumor tissues and serum of BC patients was associated with higher tumor burden, shorter survival and poorer clinical response to therapy. Conclusions: circCDYL promotes BC progression via the miR-1275-ATG7/ULK1-autophagic axis and circCDYL could act as a potential prognostic and predictive molecule for breast cancer patients.