Deep Learning for Genomics: From Early Neural Nets to Modern Large Language Models.

The data explosion driven by advancements in genomic research, such as high-throughput sequencing techniques, is constantly challenging conventional methods used in genomics. In parallel with the urgent demand for robust algorithms, deep learning has succeeded in various fields such as vision, speech, and text processing. Yet genomics entails unique challenges to deep learning, since we expect a superhuman intelligence that explores beyond our knowledge to interpret the genome from deep learning. A powerful deep learning model should rely on the insightful utilization of task-specific knowledge. In this paper, we briefly discuss the strengths of different deep learning models from a genomic perspective so as to fit each particular task with proper deep learning-based architecture, and we remark on practical considerations of developing deep learning architectures for genomics. We also provide a concise review of deep learning applications in various aspects of genomic research and point out current challenges and potential research directions for future genomics applications. We believe the collaborative use of ever-growing diverse data and the fast iteration of deep learning models will continue to contribute to the future of genomics.

Analysis of growth resistance mechanisms and causes in tea plants (Camellia sinensis) in high-pH regions of Northern China.

Background: In tea plantations with high-pH (pH > 6.5) in Northern China, tea plants are prone to yellowing disease, albinism, and reductions in components that contribute to plant quality, which affect the scale and rate of tea plantation development in Northern China. Methods: To investigate the potential causes of these issues, Camellia sinensis cv. Pingyang Tezao and Camellia sinensis cv. Ruixue were planted in Shouguang city (a high-pH area, soil pH > 6.5) and Rizhao city (a normal-pH area, soil pH is 4.5-5.5), respectively; differences in growth morphology, pigment content, cell structure, quality-determining components, and element content of the two varieties in the two areas were analyzed. Results: The results showed that tea leaves planted in Shouguang had varying degrees of yellowing disease and albinism; the pigment content in both varieties was significantly lower when planted in Shouguang compared with Rizhao. The cell structure was severely damaged and the main quality-determining components were decreased. Nitrogen (N), phosphorus (P), potassium (K), zinc (Zn), copper (Cu) and manganese (Mn) contents in the leaves of the two tea plant varieties were significantly lower when planted in Shouguang compared with those in Rizhao; the levels of these elements in Shouguang soil were significantly higher than in Rizhao soil. Calcium (Ca) contents in Shouguang soil was 9.90 times higher than that of Rizhao soil. Conclusions: We conclude that the soil in high-pH areas hindered tea plant uptake of N, Zn, Cu, and Mn, which had a detrimental effect on chloroplasts and reductions in chlorophyll synthesis, contributing to yellowing disease and albinism. In addition, excessive calcium (Ca) in Shouguang soil was also an important contributor to these negative effects. High-pH soil hindered tea plant uptake of P and K, resulting in reductions in tea polyphenols, amino acids, and other major quality components.

Automatic Plant Disease Detection Based on Tranvolution Detection Network With GAN Modules Using Leaf Images.

The detection of plant disease is of vital importance in practical agricultural production. It scrutinizes the plant's growth and health condition and guarantees the regular operation and harvest of the agricultural planting to proceed successfully. In recent decades, the maturation of computer vision technology has provided more possibilities for implementing plant disease detection. Nonetheless, detecting plant diseases is typically hindered by factors such as variations in the illuminance and weather when capturing images and the number of leaves or organs containing diseases in one image. Meanwhile, traditional deep learning-based algorithms attain multiple deficiencies in the area of this research: (1) Training models necessitate a significant investment in hardware and a large amount of data. (2) Due to their slow inference speed, models are tough to acclimate to practical production. (3) Models are unable to generalize well enough. Provided these impediments, this study suggested a Tranvolution detection network with GAN modules for plant disease detection. Foremost, a generative model was added ahead of the backbone, and GAN models were added to the attention extraction module to construct GAN modules. Afterward, the Transformer was modified and incorporated with the CNN, and then we suggested the Tranvolution architecture. Eventually, we validated the performance of different generative models' combinations. Experimental outcomes demonstrated that the proposed method satisfyingly achieved 51.7% (Precision), 48.1% (Recall), and 50.3% (mAP), respectively. Furthermore, the SAGAN model was the best in the attention extraction module, while WGAN performed best in image augmentation. Additionally, we deployed the proposed model on Hbird E203 and devised an intelligent agricultural robot to put the model into practical agricultural use.

Heat Shock Protein Member 8 (HSPA8) Is Involved in Porcine Reproductive and Respiratory Syndrome Virus Attachment and Internalization.

Porcine reproductive and respiratory syndrome virus (PRRSV), a porcine arterivirus, causes severe financial losses to global swine industry. Despite much research, the molecular mechanisms of PRRSV infection remains to be fully elucidated. In the current study, we uncovered the involvement of heat shock protein member 8 (HSPA8) in PRRSV attachment and internalization during infection for the first time. In detail, HSPA8 was identified to interact with PRRSV glycoprotein 4 (GP4), a major determinant for viral cellular tropism, dependent on its carboxy-terminal peptide-binding (PB) domain. Chemical inhibitors and specific small interference RNAs (siRNAs) targeting HSPA8 significantly suppressed PRRSV infection as indicated by decreased viral RNA abundance, infectivity, and titers. Especially, PRRSV attachment was inhibited by interference of its binding to HSPA8 with mouse anti-HSPA8 polyclonal antibodies (pAbs) and recombinant soluble HSPA8 protein. HSPA8 was further shown to participate in PRRSV internalization through clathrin-dependent endocytosis (CME). Collectively, these results demonstrate that HSPA8 is important for PRRSV attachment and internalization, which is a potential target to prevent and control the viral infection. IMPORTANCE PRRSV has caused huge economic losses to the pork industry around the world. Currently, safe and effective strategies are still urgently required to prevent and control PRRSV infection. As the first steps, PRRSV attachment and internalization are initiated by interactions between viral envelope proteins and host cell receptors/factors, which are not fully understood yet. Here, we identified the interaction between PRRSV GP4 and HSPA8, and demonstrated that HSPA8 was involved in PRRSV attachment and internalization. This work deepens our understanding of the molecular mechanisms involved in PRRSV infection, and provides novel insights for the development of antiviral drugs and vaccines against the virus.

RPLP1, an NS4B-interacting protein, enhances production of CSFV through promoting translation of viral genome.

Classical swine fever virus (CSFV), the etiological agent of classical swine fever (CSF), causes serious financial losses to the pig industry. Using yeast two-hybrid screening, we have previously identified ribosomal protein RPLP1 as a potential binding partner of CSFV NS4B. In this study, the interaction between host RPLP1 and CSFV NS4B was further characterized by co-immunoprecipitation (co-IP), glutathione S-transferase (GST) pulldown, and confocal microscopy. In addition, lentivirus-mediated shRNA knockdown of RPLP1 drastically attenuated CSFV growth, while stable overexpression of RPLP1 markedly enhanced CSFV production. Moreover, cellular RPLP1 expression was found to be significantly up-regulated along with CSFV infection. Dual-luciferase reporter assay showed that depletion of RPLP1 had no effects on the activity of CSFV internal ribosome entry site (IRES). In the first life cycle of CSFV, further studies revealed that RPLP1 depletion did not influence the intracellular viral RNA abundance but diminished the intracellular and extracellular progeny virus titers as well as the viral E2 protein expression, which indicates that RPLP1 is crucial for CSFV genome translation. In summary, this study demonstrated that RPLP1 interacts with CSFV NS4B and enhances virus production via promoting translation of viral genome.

Tumor Susceptibility Gene 101 (TSG101) Contributes to Virion Formation of Porcine Reproductive and Respiratory Syndrome Virus via Interaction with the Nucleocapsid (N) Protein along with the Early Secretory Pathway.

Porcine reproductive and respiratory syndrome virus (PRRSV) has caused huge economic losses to global swine industry. As an intracellular obligate pathogen, PRRSV exploits host cellular machinery to establish infection. The endocytic sorting complex required for transport (ESCRT) system has been shown to participate in different life cycle stages of multiple viruses. In the present study, a systematic small interference RNA screening assay identified that certain ESCRT components contributed to PRRSV infection. Among them, tumor susceptibility gene 101 (TSG101) was demonstrated to be important for PRRSV infection by knockdown and overexpression assays. TSG101 was further revealed to be involved in virion formation rather than viral attachment, internalization, RNA replication and nucleocapsid (N) protein translation within the first round of PRRSV life cycle. In detail, TSG101 was determined to specially interact with PRRSV N protein and take effect on its subcellular localization along with the early secretory pathway. Taken together, these results provide evidence that TSG101 is a proviral cellular factor for PRRSV assembly, which will be a promising target to interfere with the viral infection. IMPORTANCE PRRSV infection results in a serious swine disease affecting pig farming in the world. However, efficient prevention and control of PRRSV is hindered by its complicated infection process. Until now, our understanding of PRRSV assembly during infection is especially limited. Here, we identified that TSG101, an ESCRT-I subunit, facilitated virion formation of PRRSV via interaction with the viral N protein along with the early secretory pathway. Our work actually expands the knowledge of PRRSV infection and provides a novel therapeutic target for prevention and control of the virus.

The membrane-associated E3 ubiquitin ligase MARCH3 downregulates the IL-6 receptor and suppresses colitis-associated carcinogenesis.

The IL-6-STAT3 axis is critically involved in inflammation-associated carcinogenesis (IAC). How this axis is regulated to modulate IAC remains unknown. Here, we show that the plasma membrane-associated E3 ubiquitin ligase MARCH3 negatively regulates STAT3 activation triggered by IL-6, as well as another IL-6 subfamily member, Oncostatin M (OSM). MARCH3 is associated with the IL-6 receptor α-chain (IL-6Rα) and its coreceptor gp130. Biochemical experiments indicated that MARCH3 mediates the polyubiquitination of IL-6Rα at K401 and gp130 at K849 following IL-6 stimulation, leading to their translocation to and degradation in lysosomes. MARCH3 deficiency increases IL-6- and OSM-triggered activation of STAT3 and induction of downstream effector genes in various cell types. MARCH3 deficiency enhances dextran sulfate sodium (DSS)-induced STAT3 activation, increases the expression of inflammatory cytokines, and exacerbates colitis, as well as azoxymethane (AOM)/DSS-induced colitis-associated cancer in mice. In addition, MARCH3 is downregulated in human colorectal cancer tissues and associated with poor survival across different cancer types. Our findings suggest that MARCH3 is a pivotal negative regulator of IL-6-induced STAT3 activation, inflammation, and inflammation-associated carcinogenesis.

Preparation of docetaxel-loaded, glycyrrhetinic acid-modified nanoparticles and their liver-targeting and antitumor activity.

Liver cancer is one of the most common malignancies worldwide and poses a serious threat to human health. The most important treatment method, liver cancer chemotherapy, is limited due to its high toxicity and poor specificity. Targeted drug delivery systems have emerged as novel therapeutic strategies that deliver precise, substantial drug doses to target sites via targeting vectors and enhance the therapeutic efficacy. In the present study, glycyrrhetinic acid-modified hyaluronic acid (GA-HA) was used as a carrier for the model drug docetaxel (DTX) to prepare DTX-loaded GA-HA nanoparticles (DTX/GA-HA-NPs). The results indicated that the DTX/GA-HA-NPs exhibited high monodispersity (particle dispersity index, 0.209±0.116) and desirable particle size (208.73±5.0 nm) and zeta potential (-27.83±3.14 mV). The drug loading capacity and encapsulation efficiency of the NPs were 12.59±0.68 and 85.38±4.62%, respectively. Furthermore, it was determined that FITC-GA-HA was taken up by cells and distributed in the cytoplasm. DTX and DTX/GA-HA (just the DTX delivered by the nanoparticle) aggregated and altered the structure of cellular microtubules. Compared with DTX alone, DTX/GA-HA-NPs had a stronger inhibitory effect on HepG2 cell proliferation and promoted apoptosis of HepG2 cells. All experimental results indicated that DTX/GA-HA-NPs were successfully prepared and had liver-targeting and antitumor activities in vitro, which provided a foundation for future in vivo studies of the antitumor effects of DTX/GA-HA-NPs.

Vimentin rearrangement by phosphorylation is beneficial for porcine reproductive and respiratory syndrome virus replication in vitro.

Vimentin, a member of intermediate filaments, has been documented to be involved in viral infections. Despite several studies focusing on its involvement in porcine reproductive and respiratory syndrome virus (PRRSV) infection, the detailed mechanisms by which vimentin takes effect remain to be fully elucidated. In the present study, we identified a previously unrecognized role of vimentin rearrangement in PRRSV replication. We monitored that PRRSV infection induced vimentin reorganization during post-entry stage, which was beneficial for viral replication. In detail, the serine residue of vimentin was phosphorylated at position 38 (Ser38) by calcium calmodulin-dependent protein kinase II gamma (CaMKIIγ), and vimentin filaments reorganized into cage-like structures enwrapping PRRSV replication complex (RC) at the perinuclear location. Taken together, these results expand the knowledge of PRRSV replication, and provide novel targets for prevention and control of PRRSV.

Trapping a moving droplet train by bubble guidance in microfluidic networks.

Trapping a train of moving droplets into preset positions within a microfluidic device facilitates the long-term observation of biochemical reactions inside the droplets. In this paper, a new bubble-guided trapping method, which can remarkably improve the limited narrow two-phase flow rate range of uniform trapping, was proposed by taking advantage of the unique physical property that bubbles do not coalescence with two-phase fluids and the hydrodynamic characteristic of large flow resistance of bubbles. The flow behaviors of bubble-free and bubble-guided droplet trains were compared and analyzed under the same two-phase flow rates. The experimental results show that the droplets trapped by bubble-free guided trapping exhibit the four trapping modes of sequentially uniform trapping, non-uniform trapping induced by break-up and collision, and failed trapping due to squeezing through, and the droplets exhibit the desired uniform trapping in a relatively small two-phase flow rate range. Compared with bubble-free guided droplets, bubble-guided droplets also show four trapping modes. However, the two-phase flow rate range in which uniform trapping occurs is increased significantly and the uniformity of the trapped droplet array is improved. This investigation is beneficial to enhance the applicability of microfluidic chips for storing droplets in a passive way.

Rab1A is required for assembly of classical swine fever virus particle.

Rab1A belongs to the small Rab GTPase family and is involved in the lifecycle of numerous viruses. Here, knockdown of Rab1A inhibited CSFV growth. Further study revealed that Rab1A depletion decreased intracellular and extracellular CSFV titers, but did not affect intracellular virus genome copies and E2 protein expression within a virus lifecycle, which suggested that Rab1A is required for CSFV particle assembly rather than for genome replication or virion release. This was proofed by blocking the spread of virus using neutralizing antibodies, through which the negative effects of Rab1A knockdown on multi-cycle replication of CSFV were eliminated. Moreover, co-immunoprecipitation and confocal microscopy assays showed that Rab1A bound to CSFV NS5A protein, indicating that Rab1A and viral NS5A proteins may work cooperatively during CSFV particle assembly. In conclusion, this study demonstrated for the first time that Rab1A is required for CSFV particle assembly and binds to viral particle assembly-related NS5A protein.

Rab5 Enhances Classical Swine Fever Virus Proliferation and Interacts with Viral NS4B Protein to Facilitate Formation of NS4B Related Complex.

Classical swine fever virus (CSFV) is a fatal pig pestivirus and causes serious financial losses to the pig industry. CSFV NS4B protein is one of the most important viral replicase proteins. Rab5, a member of the small Rab GTPase family, is involved in infection and replication of numerous viruses including hepatitis C virus and dengue virus. Until now, the effects of Rab5 on the proliferation of CSFV are poorly defined. In the present study, we showed that Rab5 could enhance CSFV proliferation by utilizing lentivirus-mediated constitutive overexpression and eukaryotic plasmid transient overexpression approaches. On the other hand, lentivirus-mediated short hairpin RNA knockdown of Rab5 dramatically inhibited virus production. Co-immunoprecipitation, glutathione S-transferase pulldown and laser confocal microscopy assays further confirmed the interaction between Rab5 and CSFV NS4B protein. In addition, intracellular distribution of NS4B-Red presented many granular fluorescent signals (GFS) in CSFV infected PK-15 cells. Inhibition of basal Rab5 function with Rab5 dominant negative mutant Rab5S34N resulted in disruption of the GFS. These results indicate that Rab5 plays a critical role in facilitating the formation of the NS4B related complexes. Furthermore, it was observed that NS4B co-localized with viral NS3 and NS5A proteins in the cytoplasm, suggesting that NS3 and NS5A might be components of the NS4B related complex. Taken together, these results demonstrate that Rab5 positively modulates CSFV propagation and interacts with NS4B protein to facilitate the NS4B related complexes formation.

An Effective Approach to Improving Cadmium Telluride (111)A Surface by Molecular-Beam-Epitaxy Growth of Tellurium Monolayer.

The surface cleansing treatment of non-natural cleavage planes of semiconductors is usually performed in vacuum using ion sputtering and subsequent annealing. In this Research Article, we report on the evolution of surface atomic structure caused by different ways of surface treatment as monitored by in situ core-level photoemission measurements of Cd-4d and Te-4d atomic levels and reflection high-energy electron diffraction (RHEED). Sputtering of surface increases the density of the dangling bonds by 50%. This feature and the less than ideal ordering can be detrimental to device applications. An effective approach is employed to improve the quality of this surface. One monolayer (ML) of Te grown by the method of molecular beam epitaxy (MBE) on the target surface with heating at 300 °C effectively improves the surface quality as evidenced by the improved sharpness of RHEED pattern and a reduced diffuse background in the spectra measured by high-resolution ultraviolet photoemission spectroscopy (HRUPS). Calculations have been performed for various atomic geometries by employing first-principles geometry optimization. In conjunction with an analysis of the core level component intensities in terms the layer-attenuation model, we propose a "vacancy site" model of the modified 1 ML-Te/CdTe(111)A (2 × 2) surface.

Illuminating the surface spin texture of the giant-Rashba quantum-well system Bi/Ag(111) by circularly polarized photoemission.

We have mapped out the spin texture of a Bi/Ag surface alloy prepared on a thin Ag film by circularly polarized angle-resolved photoemission spectroscopy. A term proportional to ∇·A in the interaction Hamiltonian gives rise to strong surface photoexcitation, which interferes with a Rashba contribution to yield a pronounced circular dichroic effect in Bi/Ag. The dipole transition, often taken to be the only important photoexcitation mechanism, is actually negligible. A parameter-free calculation yields a dichroic pattern in excellent agreement with experiment.

Phonon-induced gaps in graphene and graphite observed by angle-resolved photoemission.

Mapping by angle-resolved photoemission spectroscopy of the spectral functions of graphite and graphene layers at low temperatures reveals a heretofore unreported gap of ~ 67 meV at normal emission. This gap persists to room temperature and beyond, and diminishes for increasing emission angles. We show that this gap arises from electronic coupling to out-of-plane vibrational modes at the K(¯) point in the surface Brillouin zone in accordance with conservation laws and selection rules governed by quantum mechanics. Our study suggests a new approach for characterizing phonons and electron-phonon coupling in solids.