Adaptive multi-source domain collaborative fine-tuning for transfer learning.

Fine-tuning is an important technique in transfer learning that has achieved significant success in tasks that lack training data. However, as it is difficult to extract effective features for single-source domain fine-tuning when the data distribution difference between the source and the target domain is large, we propose a transfer learning framework based on multi-source domain called adaptive multi-source domain collaborative fine-tuning (AMCF) to address this issue. AMCF utilizes multiple source domain models for collaborative fine-tuning, thereby improving the feature extraction capability of model in the target task. Specifically, AMCF employs an adaptive multi-source domain layer selection strategy to customize appropriate layer fine-tuning schemes for the target task among multiple source domain models, aiming to extract more efficient features. Furthermore, a novel multi-source domain collaborative loss function is designed to facilitate the precise extraction of target data features by each source domain model. Simultaneously, it works towards minimizing the output difference among various source domain models, thereby enhancing the adaptability of the source domain model to the target data. In order to validate the effectiveness of AMCF, it is applied to seven public visual classification datasets commonly used in transfer learning, and compared with the most widely used single-source domain fine-tuning methods. Experimental results demonstrate that, in comparison with the existing fine-tuning methods, our method not only enhances the accuracy of feature extraction in the model but also provides precise layer fine-tuning schemes for the target task, thereby significantly improving the fine-tuning performance.

Fabrication of Temperature- and Alcohol-Responsive Photonic Crystal Hydrogel and Its Application for Sustained Drug Release.

Herein, crack-free photonic crystal templates with enhanced color contrast were first demonstrated by the coassembly of polystyrene (PS) microspheres and graphene oxide (GO). Then, photonic crystal hydrogels (PCHs) with quick responses to temperature and alcohol solution concentration changes were fabricated by photopolymerization of monomers in the gaps of the self-assembled colloidal crystal templates. The structural color of the PCHs changed from yellow to blue within 120 s as the temperature rose from 25 to 40 °C, whereas upon a decrease in temperature from 40 to 25 °C, the structural color changed from blue to yellow. The structural color of the PCHs also shows an obvious response with the concentration of alcohol solution ranging from 40 to 100 wt %. The quick responses of the PCHs' structural color to changes in temperature and alcohol solution concentration are attributed to the temperature sensitivity of poly(N-isopropylacrylamide) and preferential adsorption and swelling of the alcohol solution for the polymer chains. Furthermore, moxifloxacin (Mox) was loaded into PCHs by hydrogel swelling and exhibited sustained released by increasing the temperature. The sustained release process was facilely monitored by observing the corresponding color changes in real time. The rapid and visible response offers the fabricated PCHs great potential application prospects in the semiquantitative analysis of alcohol concentration and intelligent drug delivery.

Vorasidenib (AG-881): A First-in-Class, Brain-Penetrant Dual Inhibitor of Mutant IDH1 and 2 for Treatment of Glioma.

Inhibitors of mutant isocitrate dehydrogenase (mIDH) 1 and 2 cancer-associated enzymes prevent the accumulation of the oncometabolite d-2-hydroxyglutarate (2-HG) and are under clinical investigation for the treatment of several cancers harboring an IDH mutation. Herein, we describe the discovery of vorasidenib (AG-881), a potent, oral, brain-penetrant dual inhibitor of both mIDH1 and mIDH2. X-ray cocrystal structures allowed us to characterize the compound binding site, leading to an understanding of the dual mutant inhibition. Furthermore, vorasidenib penetrates the brain of several preclinical species and inhibits 2-HG production in glioma tissue by >97% in an orthotopic glioma mouse model. Vorasidenib represents a novel dual mIDH1/2 inhibitor and is currently in clinical development for the treatment of low-grade mIDH glioma.

AG-348 enhances pyruvate kinase activity in red blood cells from patients with pyruvate kinase deficiency.

Pyruvate kinase (PK) deficiency is a rare genetic disease that causes chronic hemolytic anemia. There are currently no targeted therapies for PK deficiency. Here, we describe the identification and characterization of AG-348, an allosteric activator of PK that is currently in clinical trials for the treatment of PK deficiency. We demonstrate that AG-348 can increase the activity of wild-type and mutant PK enzymes in biochemical assays and in patient red blood cells treated ex vivo. These data illustrate the potential for AG-348 to restore the glycolytic pathway activity in patients with PK deficiency and ultimately lead to clinical benefit.

Identification of a differentially-expressed gene in fatty liver of overfeeding geese.

In response to overfeeding, geese develop fatty liver. To understand the fattening mechanism, mRNA differential display reverse transcription PCR was used to study the gene expression differences between French Landes grey geese and Xupu white geese in conditions of overfeeding and normal feeding. One gene was found to be up-regulated in the fatty liver in both breeds, and it has a 1797 bp cDNA with 83% identity to chicken SELENBP1. The sequence analysis revealed that its open reading frame of 1413 bp encodes a protein of 471 amino acids, which contains a putative conserved domain of 56 kDa selenium binding protein with high homology to its homologues of chicken (95%), rat (86%), mouse (84%), human (86%), monkey (86%), dog (86%), and cattle (86%). The function of this protein has been briefly reviewed based on published information. In tissue expression analysis, the expression of geese SELENBP1 mRNA was found to be higher in liver or kidney than in other tested tissues. The results showed that overfeeding could increase the mRNA expression level of geese SELENBP1.