Abstract Promoter design remains one of the most important considerations in metabolic engineering and synthetic biology applications. Theoretically, there are 450 possible sequences for a 50-nt promoter, which naturally occurring promoters make up only small subset. To explore vast number potential sequences, we report novel AI-based framework de novo promoter Escherichia coli. The model, was guided by sequence features learned from natural promoters, could capture interactions between...

Tongue diagnosis is a unique method in traditional Chinese medicine (TCM). This the first investigation on association between tongue and coating microbiome using next-generation sequencing. The study included 19 gastritis patients with typical white-greasy or yellow-dense corresponding to TCM Cold Hot Syndrome respectively, as well eight healthy volunteers. An Illumina paired-end, double-barcode 16S rRNA sequencing protocol was designed profile tongue-coating microbiome, from which...

The development of gastritis is associated with an increased risk gastric cancer. Current invasive diagnostic methods are not suitable for monitoring progress. In this work based on 78 patients and 50 healthy individuals, we observed that the variation tongue-coating microbiota was occurrence gastritis. Twenty-one microbial species were identified differentiating microbiomes individuals. Pathways such as metabolism in diverse environments, biosynthesis antibiotics bacterial chemotaxis...

Most weakly supervised semantic segmentation (WSSS) methods follow the pipeline that generates pseudo-masks initially and trains model with in fully manner after. However, we find some matters related to pseudo-masks, including high quality generation from class activation maps (CAMs), training noisy pseudo-mask supervision. For these matters, propose following designs push performance new state-of-art: (i) Coefficient of Variation Smoothing smooth CAMs adaptively; (ii) Proportional...

Designing promoters with desirable properties is essential in synthetic biology. Human experts are skilled at identifying strong explicit patterns small samples, while deep learning models excel detecting implicit weak large datasets. Biologists have described the sequence of via transcription factor binding sites (TFBSs). However, flanking sequences cis-regulatory elements, long been overlooked and often arbitrarily decided promoter design. To address this limitation, we introduce DeepSEED,...

In this work, we explore data augmentations for knowledge distillation on semantic segmentation. Due the capacity gap, small-sized student networks struggle to discover discriminative feature space learned by a powerful teacher. Image-level allow better imitate teacher providing extra outputs. However, existing frameworks only augment limited number of samples, which restricts learning student. Inspired recent progress directions space, work proposes feature-level augmented (FAKD) infinitely...

Bacterioplankton plays a key role in nutrient cycling and is closely related to water eutrophication algal bloom. We used high-throughput 16S rRNA gene sequencing profile archaeal bacterial community compositions the surface of Lake Taihu. It one largest lakes China has suffered from recurring cyanobacterial A total 81 samples were collected 9 different sites months 2012. found that temporal variation microbial was significantly greater than spatial (adonis, n = 9999, P < 1e-4). The...

Elastic weight consolidation (EWC) has been successfully applied for general incremental learning to overcome the catastrophic forgetting issue. It adaptively constrains each parameter of new model not deviate much from its counterpart in old during fine-tuning on class data sets, according importance tasks. However, previous study demonstrates that it still suffers when directly used object detection. In this article, we show EWC is effective detection if with critical adaptations. First,...

Current knowledge distillation approaches in semantic segmentation tend to adopt a holistic approach that treats all spatial locations equally. However, for dense prediction, students' predictions on edge regions are highly uncertain due contextual information leakage, requiring higher sensitivity than the body regions. To address this challenge, paper proposes novel called boundary-privileged (BPKD). BPKD distills of teacher model's and edges separately compact student model. Specifically,...

Abstract This paper investigates the capability of plain Vision Transformers (ViTs) for semantic segmentation using encoder–decoder framework and introduce SegViTv2 . In this study, we a novel Attention-to-Mask (ATM) module to design lightweight decoder effective ViT. The proposed ATM converts global attention map into masks high-quality results. Our outperforms popular UPerNet various ViT backbones while consuming only about $$5\%$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML">...

Enhancers serve as pivotal regulators of gene expression throughout various biological processes by interacting with transcription factors (TFs). While factor binding sites (TFBSs) are widely acknowledged key determinants TF and enhancer activity, the significant role their surrounding context sequences remains to be quantitatively characterized. Here we propose concept unit (TFBU) modularly model enhancers quantifying impact TFBSs using deep learning models. Based on this concept, develop...

Network compression has been widely studied since it is able to reduce the memory and computation cost during inference. However, previous methods seldom deal with complicated structures like residual connections, group/depth-wise convolution feature pyramid network, where channels of multiple layers are coupled need be pruned simultaneously. In this paper, we present a general channel pruning approach that can applied various structures. Particularly, propose layer grouping algorithm find...

Object detection has made enormous progress and been widely used in many applications. However, it performs poorly when only limited training data is available for novel classes that the model never seen before. Most existing approaches solve few-shot tasks implicitly without directly modeling detectors classes. In this article, we propose GenDet, a new meta-learning-based framework can effectively generate object from few shots and, thus, conducts explicitly. The detector generator trained...

Vitexicarpin (VIT) isolated from the fruits of Vitex rotundifolia has shown antitumor, anti-inflammatory, and immunoregulatory properties. This work is designed to evaluate antiangiogenic effects VIT address underlying action mechanism by a network pharmacology approach. The results validated that can act as novel angiogenesis inhibitor. Firstly, exert good inhibiting vascular-endothelial-growth-factor- (VEGF-) induced endothelial cell proliferation, migration, capillary-like tube formation...

Medical vision language pre-training (VLP) has emerged as a frontier of research, enabling zero-shot pathological recognition by comparing the query image with textual descriptions for each disease. Due to complex semantics biomedical texts, current methods struggle align medical images key findings in unstructured reports. This leads misalignment target disease's representation. In this paper, we introduce novel VLP framework designed dissect disease into their fundamental aspects,...

Here, we report an unconventional Chinese pedigree consisting of three branches all segregating prelingual hearing loss (HL) with unclear inheritance pattern. After identifying the cause one branch as maternally inherited aminoglycoside-induced HL, targeted next generation sequencing (NGS) was applied to identify genetic causes for other two branches. One affected subject from each NGS whose genomic DNA enriched either by whole-exome capture (Agilent SureSelect All Exon 50 Mb) or candidate...

Autosomal dominant types of nonsyndromic hearing loss (ADNSHL) are typically postlingual in onset and progressive. High genetic heterogeneity, late age, possible confounding due to nongenetic factors hinder the timely molecular diagnoses for most patients. In this study, exome sequencing was applied investigate a large Chinese family segregating ADNSHL which we initially failed find strong evidence linkage any locus by whole-genome analysis. Two affected members were selected sequencing. We...

Congenital absence of the uterus and vagina (CAUV) is most extreme female Müllerian duct abnormality. Several researches proposed that genetic factors contributed to this disorder, whereas precise mechanism far from full elucidation. Here, utilizing whole-exome sequencing (WES), we identified one novel missense mutation in LHX1 (NM_005568: c.G1108A, p.A370T) ten unrelated patients diagnosed with CAUV. This was absent public databases our internal database. Through luciferase reporter...

Abstract Inherited neuropathies show considerable heterogeneity in clinical manifestations and genetic etiologies are therefore often difficult to diagnose. Whole-exome sequencing (WES) has been widely adopted make definite diagnosis of unclear conditions, with proven efficacy optimizing patients’ management. In this study, a large Chinese kindred segregating autosomal dominant polyneuropathy incomplete penetrance was ascertained through patient who initially diagnosed as Charcot-Marie-Tooth...

Polycythemia vera (PV) is one of the rare causes cerebrovascular disease, whose common manifestations in nervous system are cerebral infarction and transient ischemic attack. A number cases PV patients with bleeding complicated subdural hemorrhage or have been previously reported. However, sometimes patient lower extremity venous thrombosis admitted to People's Hospital Liaoning Province. The present case study reports on a acute multiple micro-hemorrhage associated PV, who was not treated...