Slimmable transformer with hybrid axial-attention for medical image segmentation.

The transformer architecture has achieved remarkable success in medical image analysis owing to its powerful capability for capturing long-range dependencies. However, due to the lack of intrinsic inductive bias in modeling visual structural information, the transformer generally requires a large-scale pre-training schedule, limiting the clinical applications over expensive small-scale medical data. To this end, we propose a slimmable transformer to explore intrinsic inductive bias via position information for medical image segmentation. Specifically, we empirically investigate how different position encoding strategies affect the prediction quality of the region of interest (ROI) and observe that ROIs are sensitive to different position encoding strategies. Motivated by this, we present a novel Hybrid Axial-Attention (HAA) that can be equipped with pixel-level spatial structure and relative position information as inductive bias. Moreover, we introduce a gating mechanism to achieve efficient feature selection and further improve the representation quality over small-scale datasets. Experiments on LGG and COVID-19 datasets prove the superiority of our method over the baseline and previous works. Internal workflow visualization with interpretability is conducted to validate our success better; the proposed slimmable transformer has the potential to be further developed into a visual software tool for improving computer-aided lesion diagnosis and treatment planning.

Affinity binding of COVID-19 drug candidates (chloroquine/hydroxychloroquine) and serum albumin: Based on photochemistry and molecular docking.

Chloroquine (CQ) and hydroxychloroquine (HCQ) show good efficacy in the treatment of SARS-CoV-2 in the early stage, while they are no longer recommended due to their side effects. As an important drug delivery carrier, serum albumin (SA) is closely related to the efficacy of drugs. Here, the affinity behaviour of chloroquine and hydroxychloroquine with two SA were investigated through the multispectral method of biochemistry and computer simulation. The results showed that the intrinsic emission of both SA was quenched by CQ and HCQ in a spontaneous exothermic entropy reduction static process, which relied mainly on hydrogen bonding and van der Waals forces. The lower binding constants suggested weak binding between the two drugs and SA, which might lead to differences in efficacy and possibly even to varying side effects. Binding site recognition demonstrated that CQ preferred to bind to the two sites of both SA, while HCQ tended to bind to site I of SA. The results of conformational studies demonstrated that CQ and HCQ could affect the structure of both SA by slightly increasing the α-helix content of SA. Finally, we combine the results from experimental start with molecular simulations to suggest drug modifications to guide the design of drugs. This work has important implications for guiding drug design improvements to select CQ derivatives with fewer side effects for the treatment of COVID-19.

Identification of candidate reference genes for quantitative RT-PCR in Miscanthus sinensis subjected to various abiotic stresses.

Quantitative real-time PCR (qRT-PCR) has been widely used for studying gene expression at the transcript level. Its accuracy usually relies on the reference genes that are utilized for data normalization. Miscanthus sinensis, a perennial C4 grass with high biomass and strong resistance to adversities, is often utilized as a high value energy crop. However, no reliable reference genes have been investigated for normalizing gene expression for this species. In this study, 12 candidate reference genes were selected to identify their stability under five different abiotic stress treatments (drought, salt, cadmium, chromium and arsenic) by using geNorm, NormFinder, BestKeeper and RefFinder softwares. The results showed that 18S rRNA and Unigene33312 were the best reference genes under drought treatments. Unigene33312 and Unigene33024 were found to be the most stably expressed genes under salt stress and Cd stress. Moreover, Unigene33024 and PP2A were the most suitable reference genes under Cr stress and Unigene33024 and Sb09g019750 were deemed more suitable reference genes under As stress. In total, considering all the samples, Unigene33024 and PP2A were the most stable genes while ACTIN and Unigene26576 were the least stable reference genes for internal control. The expression patterns of two target genes (Cu/Zn SOD and CAT) were used to further verify those selected reference genes under different conditions. The results showed that the most and the least stable reference genes had clearly different expression patterns. This work comprehensively estimated the stability of reference genes in M. sinensis which may give insight to the reference genes selection in other tissues as well as other related varieties. These suggested reference genes would assist in further putative gene expression validation in M. sinensis.