The complex language of eukaryotic gene expression remains incompletely understood. Despite the importance suggested by many proteins variants statistically associated with human disease, nearly all such variants have unknown mechanisms, for example, protein-protein interactions (PPIs). In this study, we address this challenge using a recent machine learning advance-deep neural networks (DNNs). We aim at improving the performance of PPIs prediction and propose a method called DeepPPI (Deep neural networks for Protein-Protein Interactions prediction), which employs deep neural networks to learn effectively the representations of proteins from common protein descriptors. The experimental results indicate that DeepPPI achieves superior performance on the test data set with an Accuracy of 92.50%, Precision of 94.38%, Recall of 90.56%, Specificity of 94.49%, Matthews Correlation Coefficient of 85.08% and Area Under the Curve of 97.43%, respectively. Extensive experiments show that DeepPPI can learn useful features of proteins pairs by a layer-wise abstraction, and thus achieves better prediction performance than existing methods. The source code of our approach can be available via http://ailab.ahu.edu.cn:8087/DeepPPI/index.html .

A shortcut to adiabatic mode conversion in multimode waveguides using optical analogy of stimulated Raman adiabatic passage is investigated. The design of mode converters using the shortcut scheme is discussed. Computer-generated planar holograms are used to mimic the shaped pulses used to speed up adiabatic passage in quantum systems based on the transitionless quantum driving algorithm. The mode coupling properties are analyzed using the coupled mode theory and beam propagation simulations. We show reduced device length using the shortcut scheme as compared to the common adiabatic scheme. Modal evolution in the shortened device indeed follows the adiabatic eigenmode exactly amid the violation of adiabatic criterion.

Polarization-resolved, second harmonic generation (P-SHG) microscopy at single pixel resolution is utilized for medical diagnosis of pathological skin dermis. In analyzing the large area, pixel by pixel, second-order susceptibility of normal and pathological skin dermis, we found that P-SHG can be used to distinguish normal and dermal pathological conditions of keloid, morphea, and dermal elastolysis. Specifically, we found that the second order susceptibility tensor ratio of d(33)/d(31) for normal skins is 1.27+/-0.20, while the corresponding values for keloid, morphea, and dermal elastolysis are respectively 1.67+/-0.29, 1.79+/-0.30, and 1.75+/-0.31. We also found that the histograms of the d(33)/d(31) ratio for the pathological skins contain two peak values and are 1.5 times wider than that of the normal case, suggesting that the pathological dermal collagen fibers tend to be more structurally heterogeneous. Our work demonstrates that pixel-resolved, second-order susceptibility microscopy is effective for detecting heterogeneity in spatial distribution of collagen fibers and maybe used for future clinical diagnosis and in vivo studies of collagen pathological conditions.

In this work, we investigate the non-ablative, non-thermal photo-modification of collagen fibers by femtosecond Ti:Sa laser. The effect was induced and simultaneously registered during the repetitive laser scanning of type I collagen (rat tail and bovine Achilles' tendon), and bovine cornea. An irreversible increase in two-photon autofluorescence and a decrease in second harmonic generation intensities were associated with the collagen femtosecond laser photo-modification. Confocal spectral imaging revealed the formation of new fluorescent species. Controllable nonlinear photo-modification of collagen fibers and bovine cornea with approximately 2 microm spatial resolution was demonstrated.

We report experimental as well as theoretical investigation of the key factors that influence the relative timing jitter between hybrid synchronized ultrafast Yb and Er fiber laser systems. Experimental results show that, within the achievable synchronization range, the synchronization performance varies significantly with the relative injection timing between the 1 μm master and 1.5 μm slave pulses. This observation is in agreement with the insights obtained from the theoretical analysis, which identifies the retiming effect as a function of the initial condition of the master-slave pulse collision. By controlling the relative injection timing with a low-bandwidth intracavity feedback, relative timing jitter as low as 0.87 fs (within 1.9 MHz bandwidth) is successfully obtained.