Toward Graph Self-Supervised Learning With Contrastive Adjusted Zooming.

Graph representation learning (GRL) is critical for graph-structured data analysis. However, most of the existing graph neural networks (GNNs) heavily rely on labeling information, which is normally expensive to obtain in the real world. Although some existing works aim to effectively learn graph representations in an unsupervised manner, they suffer from certain limitations, such as the heavy reliance on monotone contrastiveness and limited scalability. To overcome the aforementioned problems, in light of the recent advancements in graph contrastive learning, we introduce a novel self-supervised GRL algorithm via graph contrastive adjusted zooming, namely, G-Zoom, to learn node representations by leveraging the proposed adjusted zooming scheme. Specifically, this mechanism enables G-Zoom to explore and extract self-supervision signals from a graph from multiple scales: micro (i.e., node level), meso (i.e., neighborhood level), and macro (i.e., subgraph level). First, we generate two augmented views of the input graph via two different graph augmentations. Then, we establish three different contrastiveness on the above three scales progressively, from node, neighboring, to subgraph level, where we maximize the agreement between graph representations across scales. While we can extract valuable clues from a given graph on the micro and macro perspectives, the neighboring-level contrastiveness offers G-Zoom the capability of a customizable option based on our adjusted zooming scheme to manually choose an optimal viewpoint that lies between the micro and macro perspectives to better understand the graph data. In addition, to make our model scalable to large graphs, we use a parallel graph diffusion approach to decouple model training from the graph size. We have conducted extensive experiments on real-world datasets, and the results demonstrate that our proposed model outperforms the state-of-the-art methods consistently.

The Structure-Function Relationship of Human Bleomycin Hydrolase: Mutation of a Cysteine Protease into a Serine Protease.

Human bleomycin hydrolase (hBH) catalyzes deamidation of the anticancer drug bleomycins (BLM). This enzyme is involved in BLM detoxification and drug resistance. Herein, we report the putative BLM-binding site and catalytic mechanism of hBH. The crystal structures and biochemical studies suggest that hBH cleaves its C-terminal residue without significant preference for the type of amino acid, and therefore can accordingly accommodate the ß-aminoalanine amide moiety of BLM for deamidation. Interestingly, hBH is capable of switching from a cysteine protease to a serine protease that is unable to cleave the secondary amide of hBH C-terminus but reacts with the primary amide of BLMs.

Highly Efficient Multiple Exciton Generation and Harvesting in Few-Layer Black Phosphorus and Heterostructure.

Multiple exciton generation (MEG) in semiconductors that yields two or more excitons by absorbing one high-energy photon has been proposed to break the Shockley-Queisser limit and boost photon-to-electron conversion efficiency. However, MEG performance in conventional bulk semiconductors or later colloidal nanocrystals is far from satisfactory. Here, we report efficient MEG in few-layer black phosphorus (BP), a direct narrow bandgap two-dimensional (2D) semiconductor with layer-tunable properties. MEG performance improves with decreasing layer number and reaches 2.09Eg threshold and 93% efficiency for two-layer BP, approaching energy conservation limit. The enhanced MEG can be attributed to strong Coulomb interaction and high density of states in 2D materials. Furthermore, MEG of BP shows negligible degradation in vertical heterostructure and multielectron can be extracted by interfacial transfer with near unity yield. These results suggest 2D semiconductors as an ideal system for next generation highly efficient light emission and charge transfer devices.

Prediction of residual/recurrent disease by HPV genotype after loop excision procedure for high-grade cervical intraepithelial neoplasia with negative margins.

AIMS: This study aims to evaluate whether the preconisation of high-risk human papillomavirus (HR-HPV) genotype and multiple HPV infection is predictive for residual/recurrent disease during the follow-up of high-grade cervical intraepithelial neoplasia (CIN) treated by loop electrosurgical excision procedure (LEEP) with negative margins. METHODS: Two hundred and thirty-six women (mean age 37; range 20-61) with CIN2/3 treated by LEEP conisation with negative margins confirmed by pathology examination of the surgical specimen were included. The cervical cells for HPV genotype testing by the polymerase chain reaction (PCR) were collected before, and 6, 12, 24 months after treatment, respectively. HPV genotype and multiple HPV infection were evaluated as possible predictors of residual/recurrent disease. RESULTS: Residual/recurrent disease was demonstrated by colposcopy-guided biopsy in 62 patients (26.3%) who underwent loop conisation with negative margins. Preoperative infection with either HPV16 (P = 0.007), HPV18 (P = 0.000), HPV33 (P = 0.001) or HPV45 (P = 0.019) was associated with higher rates of residual/recurrent lesions after conisation with negative margins. Preoperative infection with multiple HPV types was associated with the highest rate of residual/recurrent lesions compared with infection with single HPV type and HPV-negative cases (χ2 = 16.599, P < 0.001). CONCLUSIONS: Results demonstrate that the presence of HPV-16, 18, 33 and 45, as well as multiple HPV types pre-LEEP, is associated with higher rates of residual/recurrent disease after LEEP.

Prevalence of genital human papillomavirus infection and genotypes among women from Fujian province, PR China.

OBJECTIVE: The prevalence and distribution of oncogenic human papillomavirus (HPV) genotypes in women who underwent screening for cervical cancer in Fujian province, China, were assessed and the correlation of genotypes with the histological results was evaluated. STUDY DESIGN: Cervical samples were collected from 2338 women for cervical cancer screening. Participants were screened by liquid-based cytology and HPV genotyping using DNA Chip and referred to colposcopy and biopsy samples for further analyses if cytology results indicated undetermined significance (ASCUS) or higher. RESULTS: The prevalent types with cervical intraepithelial neoplasia 1 (CIN 1) were HPV-52 (22.5%), HPV-16 (11.6%) and HPV-CP8304 (10.1%). The prevalent types with cervical intraepithelial neoplasia 2 or 3 (CIN2/3) were HPV-16 (24.5%), HPV-33 (21.6%), and HPV-52 (19.6%). The prevalent types with carcinoma, including squamous cell carcinoma (SCC) and adenocarcinoma (ADCA), were HPV-16 (42.7%), HPV-18 (20.8%) and HPV-33 (12.5%). Analysis by odds ratio (OR) revealed that HPV-66, -68, and -CP8304 (HPV-CP8304: OR, 7.34; 95% confidence interval (CI)=3.73-14.46) were associated with CIN 1; HPV-52 was CIN 2/3-associated type (OR, 7.25; 95%CI=4.19-12.53); HPV-16, -18, -31, -33, -45, -53, -58, and -59 were associated with carcinoma (HPV-45: OR, 54.78; 95%CI=10.49-286.16). CONCLUSION: It was concluded that HPV infection in Chinese women in Fujian province showed higher frequencies of HPV-52 and HPV-33. However, HPV-16 was still the most common type in CIN2/3 and carcinoma. HPV-16, -18, -31, -33, -45, -53, -58 and -59 have more dominant oncogenic risk over other types in this area.