High-throughput single-cell RNA-seq data imputation and characterization with surrogate-assisted automated deep learning.

Single-cell RNA sequencing (scRNA-seq) technologies have been heavily developed to probe gene expression profiles at single-cell resolution. Deep imputation methods have been proposed to address the related computational challenges (e.g. the gene sparsity in single-cell data). In particular, the neural architectures of those deep imputation models have been proven to be critical for performance. However, deep imputation architectures are difficult to design and tune for those without rich knowledge of deep neural networks and scRNA-seq. Therefore, Surrogate-assisted Evolutionary Deep Imputation Model (SEDIM) is proposed to automatically design the architectures of deep neural networks for imputing gene expression levels in scRNA-seq data without any manual tuning. Moreover, the proposed SEDIM constructs an offline surrogate model, which can accelerate the computational efficiency of the architectural search. Comprehensive studies show that SEDIM significantly improves the imputation and clustering performance compared with other benchmark methods. In addition, we also extensively explore the performance of SEDIM in other contexts and platforms including mass cytometry and metabolic profiling in a comprehensive manner. Marker gene detection, gene ontology enrichment and pathological analysis are conducted to provide novel insights into cell-type identification and the underlying mechanisms. The source code is available at https://github.com/li-shaochuan/SEDIM.

scBGEDA: deep single-cell clustering analysis via a dual denoising autoencoder with bipartite graph ensemble clustering.

MOTIVATION: Single-cell RNA sequencing (scRNA-seq) is an increasingly popular technique for transcriptomic analysis of gene expression at the single-cell level. Cell-type clustering is the first crucial task in the analysis of scRNA-seq data that facilitates accurate identification of cell types and the study of the characteristics of their transcripts. Recently, several computational models based on a deep autoencoder and the ensemble clustering have been developed to analyze scRNA-seq data. However, current deep autoencoders are not sufficient to learn the latent representations of scRNA-seq data, and obtaining consensus partitions from these feature representations remains under-explored. RESULTS: To address this challenge, we propose a single-cell deep clustering model via a dual denoising autoencoder with bipartite graph ensemble clustering called scBGEDA, to identify specific cell populations in single-cell transcriptome profiles. First, a single-cell dual denoising autoencoder network is proposed to project the data into a compressed low-dimensional space and that can learn feature representation via explicit modeling of synergistic optimization of the zero-inflated negative binomial reconstruction loss and denoising reconstruction loss. Then, a bipartite graph ensemble clustering algorithm is designed to exploit the relationships between cells and the learned latent embedded space by means of a graph-based consensus function. Multiple comparison experiments were conducted on 20 scRNA-seq datasets from different sequencing platforms using a variety of clustering metrics. The experimental results indicated that scBGEDA outperforms other state-of-the-art methods on these datasets, and also demonstrated its scalability to large-scale scRNA-seq datasets. Moreover, scBGEDA was able to identify cell-type specific marker genes and provide functional genomic analysis by quantifying the influence of genes on cell clusters, bringing new insights into identifying cell types and characterizing the scRNA-seq data from different perspectives. AVAILABILITY AND IMPLEMENTATION: The source code of scBGEDA is available at https://github.com/wangyh082/scBGEDA. The software and the supporting data can be downloaded from https://figshare.com/articles/software/scBGEDA/19657911. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Automated exploitation of deep learning for cancer patient stratification across multiple types.

MOTIVATION: Recent frameworks based on deep learning have been developed to identify cancer subtypes from high-throughput gene expression profiles. Unfortunately, the performance of deep learning is highly dependent on its neural network architectures which are often hand-crafted with expertise in deep neural networks, meanwhile, the optimization and adjustment of the network are usually costly and time consuming. RESULTS: To address such limitations, we proposed a fully automated deep neural architecture search model for diagnosing consensus molecular subtypes from gene expression data (DNAS). The proposed model uses ant colony algorithm, one of the heuristic swarm intelligence algorithms, to search and optimize neural network architecture, and it can automatically find the optimal deep learning model architecture for cancer diagnosis in its search space. We validated DNAS on eight colorectal cancer datasets, achieving the average accuracy of 95.48%, the average specificity of 98.07%, and the average sensitivity of 96.24%, respectively. Without the loss of generality, we investigated the general applicability of DNAS further on other cancer types from different platforms including lung cancer and breast cancer, and DNAS achieved an area under the curve of 95% and 96%, respectively. In addition, we conducted gene ontology enrichment and pathological analysis to reveal interesting insights into cancer subtype identification and characterization across multiple cancer types. AVAILABILITY AND IMPLEMENTATION: The source code and data can be downloaded from https://github.com/userd113/DNAS-main. And the web server of DNAS is publicly accessible at 119.45.145.120:5001.

Prognostic and therapeutic significance of microbial cell-free DNA in plasma of people with acute decompensation of cirrhosis.

BACKGROUND & AIMS: Although the effect of bacterial infection on cirrhosis has been well-described, the effect of non-hepatotropic virus (NHV) infection is unknown. This study evaluated the genome fragments of circulating microorganisms using metagenomic next-generation sequencing (mNGS) in individuals with acute decompensation (AD) of cirrhosis, focusing on NHVs, and related the findings to clinical outcomes. METHODS: Plasma mNGS was performed in 129 individuals with AD of cirrhosis in the study cohort. Ten healthy volunteers and 20, 39, and 81 individuals with stable cirrhosis, severe sepsis and hematological malignancies, respectively, were enrolled as controls. Validation assays for human cytomegalovirus (CMV) reactivation were performed in a validation cohort (n = 58) and exploratory treatment was instituted. RESULTS: In the study cohort, 188 microorganisms were detected in 74.4% (96/129) of patients, including viruses (58.0%), bacteria (34.1%), fungi (7.4%) and chlamydia (0.5%). A NHV signature was identified in individuals with AD, and CMV was the most frequent NHV, which correlated with the clinical effect of empirical antibiotic treatment, progression to acute-on-chronic liver failure, and 90-day mortality. The NHV signature in individuals with acute-on-chronic liver failure was similar to that in those with sepsis and hematological malignancies. CMV was detected in 24.1% (14/58) of patients in the validation cohort. Of the 14 cases with detectable CMV by mNGS, nine were further validated by real-time PCR or pp65 antigenemia testing. Three patients with CMV reactivation received ganciclovir therapy in an exploratory manner and experienced clinical resolutions. CONCLUSIONS: The results of this study suggest that NHVs may play a pathogenic role in complicating the course of AD. Further validation is needed to define whether this should be incorporated into the routine management of individuals with AD of cirrhosis. IMPACT AND IMPLICATIONS: A non-hepatotropic virus (NHV) signature, which was similar to that in individuals with sepsis and hematological malignancies, was identified in individuals with acute decompensation of cirrhosis. The detected viral signature had clinical correlates, including clinical efficacy of empirical antibiotic treatment, progression to acute-on-chronic liver failure and short-term mortality. Cytomegalovirus reactivation, which is treatable, may adversely affect clinical outcomes in some individuals with decompensated cirrhosis. Routine screening for NHVs, especially cytomegalovirus, may be useful for the management of individuals with acute decompensation of cirrhosis.

Identification of haploinsufficient genes from epigenomic data using deep forest.

Haploinsufficiency, wherein a single allele is not enough to maintain normal functions, can lead to many diseases including cancers and neurodevelopmental disorders. Recently, computational methods for identifying haploinsufficiency have been developed. However, most of those computational methods suffer from study bias, experimental noise and instability, resulting in unsatisfactory identification of haploinsufficient genes. To address those challenges, we propose a deep forest model, called HaForest, to identify haploinsufficient genes. The multiscale scanning is proposed to extract local contextual representations from input features under Linear Discriminant Analysis. After that, the cascade forest structure is applied to obtain the concatenated features directly by integrating decision-tree-based forests. Meanwhile, to exploit the complex dependency structure among haploinsufficient genes, the LightGBM library is embedded into HaForest to reveal the highly expressive features. To validate the effectiveness of our method, we compared it to several computational methods and four deep learning algorithms on five epigenomic data sets. The results reveal that HaForest achieves superior performance over the other algorithms, demonstrating its unique and complementary performance in identifying haploinsufficient genes. The standalone tool is available at https://github.com/yangyn533/HaForest.

Identification of pan-cancer Ras pathway activation with deep learning.

The identification of hidden responders is often an essential challenge in precision oncology. A recent attempt based on machine learning has been proposed for classifying aberrant pathway activity from multiomic cancer data. However, we note several critical limitations there, such as high-dimensionality, data sparsity and model performance. Given the central importance and broad impact of precision oncology, we propose nature-inspired deep Ras activation pan-cancer (NatDRAP), a deep neural network (DNN) model, to address those restrictions for the identification of hidden responders. In this study, we develop the nature-inspired deep learning model that integrates bulk RNA sequencing, copy number and mutation data from PanCanAltas to detect pan-cancer Ras pathway activation. In NatDRAP, we propose to synergize the nature-inspired artificial bee colony algorithm with different gradient-based optimizers in one framework for optimizing DNNs in a collaborative manner. Multiple experiments were conducted on 33 different cancer types across PanCanAtlas. The experimental results demonstrate that the proposed NatDRAP can provide superior performance over other benchmark methods with strong robustness towards diagnosing RAS aberrant pathway activity across different cancer types. In addition, gene ontology enrichment and pathological analysis are conducted to reveal novel insights into the RAS aberrant pathway activity identification and characterization. NatDRAP is written in Python and available at https://github.com/lixt314/NatDRAP1.

Combinatorial Self-Assembly of Coordination Cages with Systematically Fine-Tuned Cavities for Efficient Co-Encapsulation and Catalysis.

Controllable arrangement of different ligands in a single assembly will not only bring increased complexity but also offers a new route to fine-tune the function of the designed architecture. We report here a combinatorial self-assembly with enPd(NO3 )2 and three different ligands (L1-3 ), which gave rise to a family of six palladium-organic cages (C1-6) with systematically varied shapes and cavities, including three new heteroleptic (Pd5 L1 2 L2 , Pd5 L1 2 L3 , Pd4 L2 L3 ), one new homoleptic (Pd4 L3 2 ) cages, and two known homoleptic (Pd6 L1 4 , Pd4 L2 2 ) cages. Emergent functions due to the fusion of two half cavities on the heteroleptic cages from their parent homoleptic cages have been observed: the heteroleptic cages can form ternary complexes by co-encapsulation of both aromatic and aliphatic guests, while their homoleptic counterparts can only form binary complexes. Such a forced co-encapsulation effect endows the heteroleptic cages with enhanced catalytic power for the Knoevenagel condensation.

Controlled Self-Assembly and Multistimuli-Responsive Interconversions of Three Conjoined Twin-Cages.

Stimuli-responsive structural transformations between discrete coordination supramolecular architectures not only are essential to construct smart functional materials but also provide a versatile molecular-level platform to mimic the biological transformation process. We report here the controlled self-assembly of three topologically unprecedented conjoined twin-cages, i.e., one stapled interlocked Pd12L6 cage (2) and two helically isomeric Pd6L3 cages (3 and 4) made from the same cis-blocked palladium corners and a new bis-bidentate ligand (1). While cage 2 features three mechanically coupled cavities, cages 3 and 4 are topologically isomeric helicate-based twin-cages based on the same metal/ligand stoichiometry. Sole formation of cage 2 or a dynamic mixture of cages 3 and 4 can be controlled by changing the solvents employed during the self-assembly. Structural conversions between cages 3 and 4 can be triggered by changes in both temperature/solvent and induced-fit guest encapsulations. Well-controlled interconversion between such topologically complex superstructures may lay a solid foundation for achieving a variety of functions within a switchable system.

Aberrant gut microbiota alters host metabolome and impacts renal failure in humans and rodents.

OBJECTIVE: Patients with renal failure suffer from symptoms caused by uraemic toxins, possibly of gut microbial origin, as deduced from studies in animals. The aim of the study is to characterise relationships between the intestinal microbiome composition, uraemic toxins and renal failure symptoms in human end-stage renal disease (ESRD). DESIGN: Characterisation of gut microbiome, serum and faecal metabolome and human phenotypes in a cohort of 223 patients with ESRD and 69 healthy controls. Multidimensional data integration to reveal links between these datasets and the use of chronic kidney disease (CKD) rodent models to test the effects of intestinal microbiome on toxin accumulation and disease severity. RESULTS: A group of microbial species enriched in ESRD correlates tightly to patient clinical variables and encode functions involved in toxin and secondary bile acids synthesis; the relative abundance of the microbial functions correlates with the serum or faecal concentrations of these metabolites. Microbiota from patients transplanted to renal injured germ-free mice or antibiotic-treated rats induce higher production of serum uraemic toxins and aggravated renal fibrosis and oxidative stress more than microbiota from controls. Two of the species, Eggerthella lenta and Fusobacterium nucleatum, increase uraemic toxins production and promote renal disease development in a CKD rat model. A probiotic Bifidobacterium animalis decreases abundance of these species, reduces levels of toxins and the severity of the disease in rats. CONCLUSION: Aberrant gut microbiota in patients with ESRD sculpts a detrimental metabolome aggravating clinical outcomes, suggesting that the gut microbiota will be a promising target for diminishing uraemic toxicity in those patients. TRIAL REGISTRATION NUMBER: This study was registered at ClinicalTrials.gov (NCT03010696).

Protective mechanism of SIRT1 on Hcy-induced atrial fibrosis mediated by TRPC3.

High plasma levels of homocysteine (Hcy) are regarded as a risk factor for atrial fibrillation (AF), which is closely associated with the pathological consequence of atrial fibrosis and can lead to heart failure with a high mortality rate; here, we show that atrial fibrosis is mediated by the relationship between canonical transient receptor potential 3 (TRPC3) channels and sirtuin type 1 (SIRT1) under the stimulation of Hcy. The left atrial appendage was obtained from patients with either sinus rhythm (SR) or AF and used to evaluate the relationship between the concentration of Hcy and a potential mechanism of cardiac fibrosis mediated by TRPC3 and SIRT1. We next performed transverse aortic constriction (TAC) in mouse to investigate the relationship. The mechanisms underlying atrial fibrosis involving TRPC3 and SIRT1 proteins were explored by co-IP, BLI and lentivirus transfection experiments. qPCR and WB were performed to analyse gene and protein expression, respectively. The higher level of atrial fibrosis was observed in the HH mouse group with a high Hcy diet. Such results suggest that AF patients may be more susceptible to atrial fibrosis and possess a high probability of progressing to hyperhomocysteinemia. Moreover, our findings are consistent with the hypothesis that TRPC3 channel up-regulation leads to abnormal accumulation of collagen, with the down-regulation of SIRT1 as an aetiological factor of high Hcy, which in turn predisposes to atrial fibrosis and strongly enhances the possibility of AF.

Guest-Reaction Driven Cage to Conjoined Twin-Cage Mitosis-Like Host Transformation.

We report here a guest-reaction-induced mitosis-like host transformation from a known Pd4 L2 cage 1 to a conjoined Pd6 L3 twin-cage 2 featuring two separate cavities. The encapsulation of 1-hydroxymethyl-2-naphthol (G1), a known ortho-quinone methide (o-QMs) precursor, within the hydrophobic cavity of cage 1 is found crucial to realize the cage to twin-cage conversion. Confined G1 molecules within the nanocavity undergo self-coupling dimerization reaction to form 2,2'-dihydroxy-1,1'-dinaphthylmethane (G2) which then triggers the cage to twin-cage mitosis. The same conversion also proceeds, in a much faster rate, via the direct templation of G2, confirming the induced-fit transformation mechanism. The structure of the (G2)2 ⊂2 host-guest complex has been established by X-ray crystallographic study, where cis- to trans- conformational switch on one bridging ligand is revealed.

FGF23 regulates atrial fibrosis in atrial fibrillation by mediating the STAT3 and SMAD3 pathways.

High fibroblast growth factor 23 (FGF23) concentrations are a strong predictor of atrial fibrillation (AF), but researchers have not clearly determined the mechanism by which FGF23 causes atrial fibrosis in patients with AF. This study aims to elucidate the mechanism by which FGF23 induces atrial fibrosis in patients with AF. Immunohistochemistry was used to study the expression of FGF23, FGFR4, and fibrotic factors in patients with a normal sinus rhythm (SR) and patients with AF. Cardiac fibroblasts (CFs) were cocultured with different concentrations of the recombinant FGF23 protein. Compared with the SR group, the levels of FGF23, FGFR4, α-smooth muscle actin (α-SMA), and collagen-1 were significantly increased in the AF group. Exposure to high concentrations of the recombinant FGF23 protein increased the accumulation of reactive oxygen species (ROS) and activated α-SMA, collagen-1, signal transducer and activator of transcription 3 (STAT3) and SMAD3 signaling in cultured CFs. The levels of fibrotic proteins in CFs stimulated with high concentrations of the recombinant FGF23 protein were reversed by N-acetylcysteine (NAC, a ROS inhibitor), ship information system 3 (a SMAD3 inhibitor), and Stattic (a STAT3 inhibitor). Furthermore, compared to untreated CFs, CFs treated with the recombinant FGF23 protein were characterized by an increased interaction between STAT3 and SMAD3. Based on these results, FGF23 induces atrial fibrosis in patients with AF by increasing ROS production and subsequently activating STAT3 and SMAD3 signaling.

Water-Soluble Redox-Active Cage Hosting Polyoxometalates for Selective Desulfurization Catalysis.

Transformations within container-molecules provide a good alternative between traditional homogeneous and heterogeneous catalysis, as the containers themselves can be regarded as single molecular nanomicelles. We report here the designed-synthesis of a water-soluble redox-active supramolecular Pd4L2 cage and its application in the encapsulation of aromatic molecules and polyoxometalates (POMs) catalysts. Compared to the previous known Pd6L4 cage, our results show that replacement of two cis-blocked palladium corners with p-xylene bridges through pyridinium bonds formation between the 2,4,6-tri-4-pyridyl-1,3,5-triazine (TPT) ligands not only provides reversible redox-activities for the new Pd4L2 cage, but also realizes the expansion and subdivision of its internal cavity. An increased number of guests, including polyaromatics and POMs, can be accommodated inside the Pd4L2 cage. Moreover, both conversion and product selectivity (sulfoxide over sulfone) have also been much enhanced in the desulfurization reactions catalyzed by the POMs@Pd4L2 host-guest complexes. We expect that further photochromic or photoredox functions are possible taking advantage of this new generation of organo-palladium cage.

A human gut microbial gene catalogue established by metagenomic sequencing.

To understand the impact of gut microbes on human health and well-being it is crucial to assess their genetic potential. Here we describe the Illumina-based metagenomic sequencing, assembly and characterization of 3.3 million non-redundant microbial genes, derived from 576.7 gigabases of sequence, from faecal samples of 124 European individuals. The gene set, approximately 150 times larger than the human gene complement, contains an overwhelming majority of the prevalent (more frequent) microbial genes of the cohort and probably includes a large proportion of the prevalent human intestinal microbial genes. The genes are largely shared among individuals of the cohort. Over 99% of the genes are bacterial, indicating that the entire cohort harbours between 1,000 and 1,150 prevalent bacterial species and each individual at least 160 such species, which are also largely shared. We define and describe the minimal gut metagenome and the minimal gut bacterial genome in terms of functions present in all individuals and most bacteria, respectively.

The diploid genome sequence of an Asian individual.

Here we present the first diploid genome sequence of an Asian individual. The genome was sequenced to 36-fold average coverage using massively parallel sequencing technology. We aligned the short reads onto the NCBI human reference genome to 99.97% coverage, and guided by the reference genome, we used uniquely mapped reads to assemble a high-quality consensus sequence for 92% of the Asian individual's genome. We identified approximately 3 million single-nucleotide polymorphisms (SNPs) inside this region, of which 13.6% were not in the dbSNP database. Genotyping analysis showed that SNP identification had high accuracy and consistency, indicating the high sequence quality of this assembly. We also carried out heterozygote phasing and haplotype prediction against HapMap CHB and JPT haplotypes (Chinese and Japanese, respectively), sequence comparison with the two available individual genomes (J. D. Watson and J. C. Venter), and structural variation identification. These variations were considered for their potential biological impact. Our sequence data and analyses demonstrate the potential usefulness of next-generation sequencing technologies for personal genomics.

Functional cellulose-derived epoxy cross-linked with BADGE resin to construct high-performance epoxy composites.

The development of equipped bio-based epoxy materials has been gaining much attention recently. Nevertheless, finding the balance between the structure and properties of materials remains a significant challenge. In this work, cellulose-based epoxy (PHPCEP) with "soft" and "hard" cooperative structures was designed and demonstrated to endow bisphenol A diglycidyl ether (BADGE) with excellent toughness, heat resistance, mechanical strength, glass transition temperature, thermal stability, and solvent resistance. When 5 wt% PHPCEP was incorporated into BADGE composites, the resulting materials exhibited the maximum flexural strength (121.9 MPa) and tensile strength (71.4 MPa), a high glass transition temperature (148.3 °C), and 10 wt% PHPCEP/BADGE demonstrated the highest impact strength (70.5 kJ/m2). These figures are 18.8 %, 16.1 %, 21.5 %, and 254.3 % higher than the corresponding values of neat BADGE. The results of dynamic mechanical properties and heat degradation of the cured specimens also suggest that PHPCEP/BADGE materials have superior stiffness and toughness than neat BADGE, which could be attributed to the strong interaction between PHPCEP and BADGE, delivering better thermal stability for the composites compared to the pristine resin. Considering the remarkable effect, this work provides an effective way of highly efficient utilization of abundant cellulose and a high-performance additive for composite materials.

Stepwise Synthesis of Low-Symmetry Hexacationic Pyridinium Organic Cages.

An efficient approach was developed for the synthesis of the well-known BlueCage by pre-bridging two 2,4,6-tris(4-pyridyl)-1,3,5-triazine (TPT) panels with one linker followed by cage formation in a much improved yield and shortened reaction time. Such a stepwise methodology was further applied to synthesize three new pyridinium organic cages, C2, C3, and C4, where the low-symmetry cages C3 and C4 with angled panels demonstrated better recognition properties toward 1,1'-bi-2-naphthol (BINOL) than the high-symmetry analogue C2 featuring parallel platforms.

Relationships among the gut microbiome, brain networks, and symptom severity in schizophrenia patients: A mediation analysis.

The microbiome-gut-brain axis (MGBA) plays a critical role in schizophrenia (SZ). However, the underlying mechanisms of the interactions among the gut microbiome, brain networks, and symptom severity in SZ patients remain largely unknown. Fecal samples, structural and functional magnetic resonance imaging (MRI) data, and Positive and Negative Syndrome Scale (PANSS) scores were collected from 38 SZ patients and 38 normal controls, respectively. The data of 16S rRNA gene sequencing were used to analyze the abundance of gut microbiome and the analysis of human brain networks was applied to compute the nodal properties of 90 brain regions. A total of 1,691,280 mediation models were constructed based on 261 gut bacterial, 810 nodal properties, and 4 PANSS scores in SZ patients. A strong correlation between the gut microbiome and brain networks (r = 0.89, false discovery rate (FDR) -corrected p < 0.05) was identified. Importantly, the PANSS scores were linearly correlated with both the gut microbiome (r = 0.5, FDR-corrected p < 0.05) and brain networks (r = 0.59, FDR-corrected p < 0.05). The abundance of genus Sellimonas significantly affected the PANSS negative scores of SZ patients via the betweenness centrality of white matter networks in the inferior frontal gyrus and amygdala. Moreover, 19 significant mediation models demonstrated that the nodal properties of 7 brain regions, predominately from the systems of visual, language, and control of action, showed significant mediating effects on the PANSS scores with the gut microbiome as mediators. Together, our findings indicated the tripartite relationships among the gut microbiome, brain networks, and PANSS scores and suggested their potential role in the neuropathology of SZ.

Colorectal cancer subtype identification from differential gene expression levels using minimalist deep learning.

BACKGROUND: Cancer molecular subtyping plays a critical role in individualized patient treatment. In previous studies, high-throughput gene expression signature-based methods have been proposed to identify cancer subtypes. Unfortunately, the existing ones suffer from the curse of dimensionality, data sparsity, and computational deficiency. METHODS: To address those problems, we propose a computational framework for colorectal cancer subtyping without any exploitation in model complexity and generality. A supervised learning framework based on deep learning (DeepCSD) is proposed to identify cancer subtypes. Specifically, based on the differentially expressed genes under cancer consensus molecular subtyping, we design a minimalist feed-forward neural network to capture the distinct molecular features in different cancer subtypes. To mitigate the overfitting phenomenon of deep learning as much as possible, L1 and L2 regularization and dropout layers are added. RESULTS: For demonstrating the effectiveness of DeepCSD, we compared it with other methods including Random Forest (RF), Deep forest (gcForest), support vector machine (SVM), XGBoost, and DeepCC on eight independent colorectal cancer datasets. The results reflect that DeepCSD can achieve superior performance over other algorithms. In addition, gene ontology enrichment and pathology analysis are conducted to reveal novel insights into the cancer subtype identification and characterization mechanisms. CONCLUSIONS: DeepCSD considers all subtype-specific genes as input, which is pathologically necessary for its completeness. At the same time, DeepCSD shows remarkable robustness in handling cross-platform gene expression data, achieving similar performance on both training and test data without significant model overfitting or exploitation of model complexity.

Study of Surface Integrity of Titanium Alloy (TC4) by Belt Grinding to Achieve the Same Surface Roughness Range.

Titanium alloy materials are used in a variety of engineering applications in the aerospace, aircraft, electronics, and shipbuilding industries, and due to the continuous improvement of the contemporary age, surface integrity needs to be improved for engineering applications. Belt grinding parameters and levels directly affect the surface integrity of titanium alloys (TC4), which further affects the fatigue life of the titanium alloys during service. In order to investigate the surface integrity of titanium alloys at different roughness levels, the surfaces were repeatedly ground with the same type and different models of abrasive belts. The results showed that at roughness Ra levels of 0.4 µm to 0.2 µm, the compressive residual stresses decreased with increasing linear velocity and there were problems with large surface morphological defects. At the roughness Ra of 0.2 µm or less, grinding improves the surface morphology, the compressive residual stress increases with increasing feed rate, and the surface hardness decreases with increasing linear velocity. In addition, the research facilitates the engineering of grinding parameters and levels that affect surface integrity under different roughness conditions, providing a theoretical basis and practical reference.