Single-cell type annotation with deep learning in 265 cell types for humans.

Motivation: Annotating cell types is a challenging yet essential task in analyzing single-cell RNA sequencing data. However, due to the lack of a gold standard, it is difficult to evaluate the algorithms fairly and an overfitting algorithm may be favored in benchmarks. To address this challenge, we developed a deep learning-based single-cell type prediction tool that assigns the cell type to 265 different cell types for humans, based on data from approximately five million cells. Results: We achieved a median area under the ROC curve (AUC) of 0.93 when evaluated across datasets. We found that inconsistent labeling in the existing database generated by different labs contributed to the mistakes of the model. Therefore, we used cell ontology to correct the annotations and retrained the model, which resulted in 0.971 median AUC. Our study reveals a limiting factor of the accuracy one may achieve with the current database annotation and points to the solutions towards an algorithm-based correction of the gold standard for future automated cell annotation approaches. Availability and implementation: The code is available at: https://github.com/SherrySDong/Hierarchical-Correction-Improves-Automated-Single-cell-Type-Annotation. Data used in this study are listed in Supplementary Table S1 and are retrievable at the CZI database.

Single-cell transcriptomic sequencing identifies subcutaneous patient-derived xenograft recapitulated medulloblastoma.

BACKGROUND: Medulloblastoma (MB) is one of the most common malignant brain tumors that mainly affect children. Various approaches have been used to model MB to facilitate investigating tumorigenesis. This study aims to compare the recapitulation of MB between subcutaneous patient-derived xenograft (sPDX), intracranial patient-derived xenograft (iPDX), and genetically engineered mouse models (GEMM) at the single-cell level. METHODS: We obtained primary human sonic hedgehog (SHH) and group 3 (G3) MB samples from six patients. For each patient specimen, we developed two sPDX and iPDX models, respectively. Three Patch+/- GEMM models were also included for sequencing. Single-cell RNA sequencing was performed to compare gene expression profiles, cellular composition, and functional pathway enrichment. Bulk RNA-seq deconvolution was performed to compare cellular composition across models and human samples. RESULTS: Our results showed that the sPDX tumor model demonstrated the highest correlation to the overall transcriptomic profiles of primary human tumors at the single-cell level within the SHH and G3 subgroups, followed by the GEMM model and iPDX. The GEMM tumor model was able to recapitulate all subpopulations of tumor microenvironment (TME) cells that can be clustered in human SHH tumors, including a higher proportion of tumor-associated astrocytes and immune cells, and an additional cluster of vascular endothelia when compared to human SHH tumors. CONCLUSIONS: This study was the first to compare experimental models for MB at the single-cell level, providing value insights into model selection for different research purposes. sPDX and iPDX are suitable for drug testing and personalized therapy screenings, whereas GEMM models are valuable for investigating the interaction between tumor and TME cells.

Predicting Single Neuron Responses of the Primary Visual Cortex with Deep Learning Model.

Modeling neuron responses to stimuli can shed light on next-generation technologies such as brain-chip interfaces. Furthermore, high-performing models can serve to help formulate hypotheses and reveal the mechanisms underlying neural responses. Here the state-of-the-art computational model is presented for predicting single neuron responses to natural stimuli in the primary visual cortex (V1) of mice. The algorithm incorporates object positions and assembles multiple models with different train-validation data, resulting in a 15%-30% improvement over the existing models in cross-subject predictions and ranking first in the SENSORIUM 2022 Challenge, which benchmarks methods for neuron-specific prediction based on thousands of images. Importantly, The model reveals evidence that the spatial organizations of V1 are conserved across mice. This model will serve as an important noninvasive tool for understanding and utilizing the response patterns of primary visual cortex neurons.

Xanthogranulomatous cystitis and bilateral ureteral reflux due to bladder neck obstruction: a case report.

Xanthogranulomatous cystitis (XC) is a rare benign disease of chronic granulomatous inflammation. We report a 23-year-old woman with xanthogranulomatous cystitis. She was referred to our hospital with the chief complaint of a 1-year history of frequent, urgent dysuria with recurrent fever. An imaging examination showed bilateral ureteral reflux and a normal bladder. Urodynamic findings suggested bladder outlet obstruction and increased post-void residual urine. Finally, the patient underwent endoscopy, and bladder neck obstruction was confirmed. Additionally, we found multiple granulomatous masses in the bladder. Therefore, we performed transurethral resection of the tumor and bladder neck. A histopathological examination of resected tumor tissue showed xanthogranulomatous cystitis, and the patient received anti-infective therapy. Follow-up cystourethroscopic results and urination symptoms returned to normal, and the bilateral ureteral reflux was gradually reduced.

Staphylococcal enterotoxin C as a novel strategy for treating lumbar spondylolysis in adolescents: Description of technique.

Lumbar spondylolysis is one of the most common causes of low back pain and primarily affects children and adolescents. Traditional posterior lumbar fixation and interbody fusion surgery has always been the most effective method to treat spondylolysis. However, traditional surgical management has limitations of large trauma, complex operation, high cost, postoperative biomechanical deterioration, and resulting complications. In order to avoid the trauma and complications of surgical treatment, and reduce the cost of treatment. Based on the successful clinical experience of using staphylococcal enterotoxin C (SEC) to treat nonunion after a limb fracture, we identified a minimally invasive method to effectively treat lumbar spondylolysis. A novel minimally invasive therapeutic approach is presented herein of an SEC injection guided by C-arm fluoroscopy to treat lumbar spondylolysis. We describe a novel technique applied in a patient with lumbar spondylolysis, who showed significantly improved low back pain symptoms and a computed tomography scan, including osseous fusion of the bilateral isthmus at L4 after SEC therapy. This is the first reported case description of using an SEC injection to treat lumbar spondylolysis with a successful clinical outcome.

Facile Ball Milling Preparation of Flame-Retardant Polymer Materials: An Overview.

To meet the growing needs of public safety and sustainable development, it is highly desirable to develop flame-retardant polymer materials using a facile and low-cost method. Although conventional solution chemical synthesis has proven to be an efficient way of developing flame retardants, it often requires organic solvents and a complicated separation process. In this review, we summarize the progress made in utilizing simple ball milling (an important type of mechanochemical approach) to fabricate flame retardants and flame-retardant polymer composites. To elaborate, we first present a basic introduction to ball milling, and its crushing, exfoliating, modifying, and reacting actions, as used in the development of high-performance flame retardants. Then, we report the mixing action of ball milling, as used in the preparation of flame-retardant polymer composites, especially in the formation of multifunctional segregated structures. Hopefully, this review will provide a reference for the study of developing flame-retardant polymer materials in a facile and feasible way.

Enabling Eating Detection in a Free-living Environment: Integrative Engineering and Machine Learning Study.

BACKGROUND: Monitoring eating is central to the care of many conditions such as diabetes, eating disorders, heart diseases, and dementia. However, automatic tracking of eating in a free-living environment remains a challenge because of the lack of a mature system and large-scale, reliable training set. OBJECTIVE: This study aims to fill in this gap by an integrative engineering and machine learning effort and conducting a large-scale study in terms of monitoring hours on wearable-based eating detection. METHODS: This prospective, longitudinal, passively collected study, covering 3828 hours of records, was made possible by programming a digital system that streams diary, accelerometer, and gyroscope data from Apple Watches to iPhones and then transfers the data to the cloud. RESULTS: On the basis of this data collection, we developed deep learning models leveraging spatial and time augmentation and inferring eating at an area under the curve (AUC) of 0.825 within 5 minutes in the general population. In addition, the longitudinal follow-up of the study design encouraged us to develop personalized models that detect eating behavior at an AUC of 0.872. When aggregated to individual meals, the AUC is 0.951. We then prospectively collected an independent validation cohort in a different season of the year and validated the robustness of the models (0.941 for meal-level aggregation). CONCLUSIONS: The accuracy of this model and the data streaming platform promises immediate deployment for monitoring eating in applications such as diabetic integrative care.

Heterogeneous digital biomarker integration out-performs patient self-reports in predicting Parkinson's disease.

Parkinson's disease (PD) is one of the first diseases where digital biomarkers demonstrated excellent performance in differentiating disease from healthy individuals. However, no study has systematically compared and leveraged multiple types of digital biomarkers to predict PD. Particularly, machine learning works on the fine-motor skills of PD are limited. Here, we developed deep learning methods that achieved an AUC (Area Under the receiver operator characteristic Curve) of 0.933 in identifying PD patients on 6418 individuals using 75048 tapping accelerometer and position records. Performance of tapping is superior to gait/rest and voice-based models obtained from the same benchmark population. Assembling the three models achieved a higher AUC of 0.944. Notably, the models not only correlated strongly to, but also performed better than patient self-reported symptom scores in diagnosing PD. This study demonstrates the complementary predictive power of tapping, gait/rest and voice data and establishes integrative deep learning-based models for identifying PD.

Synergistic denitrification, partial nitrification - Anammox in a novel A2/O2 reactor for efficient nitrogen removal from low C/N wastewater.

A biofilm-based anaerobic-aerobic (A2O2) reactor was constructed to treat manure-free piggery wastewater. The reactor contained four compartments, among which the first two were anaerobic (A phase) and the last two were aerobic (O phase). Throughout around one-year operation, high-level nutrient removal was demonstrated. At an optimal reflux ratio of 100%, the average NH4+-N, TN, and COD removal efficiencies were high as 99.4%, 91.7%, and 79.4%, respectively, with the influent concentration of 220.6, 231.6 and 332 mg/L, respectively. The NH4+-N, TN, and COD concentrations in the final effluent were only 1.4, 18.5 and 65 mg/L, respectively. COD and nitrogen removal were mainly removed in the A phase and O phase, respectively. This result revolutionizes the previous perception that nitrogen is only removed in the A phase of conventional A-O configuration. Achievement of PN/A in the O phase was critical to the efficient nitrogen removal. Heterotrophic denitrification in the anaerobic compartments removed the nitrate produced by anammox, ensuring the high-level nitrogen removal. Anaerobic organic degradation was a major pathway for COD removal, as abundant methanogens detected in the A phase. This study provides a feasible technical scheme for the efficient nutrient removal from ammonium-rich wastewater.

The Effect of MaxiK Channel on Regulating the Activation of NLRP3 Inflammasome in Rats of Blast-induced Traumatic Brain Injury.

Abundant findings including our previous work proved that the NOD-like receptor family, pyrin domain-containing 3 (NLRP3) inflammasome exerts a key role in the process of neuroinflammation following blast-induced traumatic brain injury (bTBI). The opening of potassium channels leads to low K+ environment in cells, which appears to be an essential requirement for NLRP3 inflammasome activation. Notably, MaxiK (BK) channel is significant for K+ transport. The present study is aim to investigate the potential role of MaxiK in the activation of NLRP3 and to evaluate whether MaxiK channel blocker paxilline could confer beneficial effects on attenuating the severity of bTBI in rats. Rats were randomly assigned into five groups (n = 8). MaxiK channel expression was measured in bTBI rats. The effect of paxilline on the expression of NLRP3 inflammasome, the level of inflammatory cytokines, brain injury biomarkers in serum and brain edema were also evaluated in bTBI rats. The results showed that the expression of MaxiK was elevated significantly in the cerebral cortex of bTBI rats. The treatment of MaxiK channel blocker paxilline suppressed the NLRP3 inflammasome expression substantially. In addition, paxilline could also decrease the level of pro-inflammatory cytokines and the biomarkers of brain injury and alleviate brain edema of bTBI rats. Our findings have revealed that MaxiK channel might be involved in the process of neuroinflammation of bTBI. Paxilline could depress neuro-inflammation response and alleviate brain injury by blocking MaxiK channel and subsequently inhibition of NLRP3 inflammasome activation.

Long non-coding RNA MIAT is involved in the regulation of pyroptosis in diabetic cardiomyopathy via targeting miR-214-3p.

Diabetic cardiomyopathy (DCM) is one of the most common complications of diabetes without effective treatment options. Its pathogenesis is complex and remains unclear. Long non-coding RNA (lncRNA) MIAT allele has been reported to be enriched in DCM patients and activate a pyroptosis program in hypoxia-induced H9c2 cells. Thus, whether MIAT played a role in DCM pyroptosis remains to be clarified. In the study, the expression of MIAT was found elevated in the serum of diabetic patients, as well as in high-glucose induced cardiomyocytes and diabetic mice. Further, the expression levels of CASP1 and pyroptosis correlation factors (IL-1 and IL-18) were downregulated after silencing MIAT. Through modeling and validation experiments, we then confirmed that the MIAT-miR-214-3p-CASP1 axis serves as an essential point in pyroptosis of DCM mice. These results suggested that silencing MIAT would be a potential treatment strategy for DCM.

Efficacy and safety of tyrosine kinase inhibitors alone or in combination with radiation therapy for metastatic renal cell carcinoma: A protocol for evidence-based systematic review and Bayesian network meta-analysis.

BACKGROUND: Given the lack of evidence for survival benefit in patients with metastatic renal cell carcinoma from the addition of radiation therapy to tyrosine kinase inhibitor therapy, this Bayesian network meta-analysis aimed to evaluate survival outcomes in patients receiving radiation therapy plus tyrosine kinase inhibitor therapy. METHODS: The preferred reporting items for systematic reviews and meta-analyses reporting guidelines were followed to conduct this study. The electronic databases of PubMed, Cochrane Library, EMBASE, and Web of Science were searched from the inception to August 2021. All phase III clinical trials that reported the outcomes of tyrosine kinase inhibitor with radiation therapy compared with those of tyrosine kinase inhibitor or radiation therapy alone for patients with metastatic renal cell carcinoma were considered eligible for inclusion in this meta-analysis. Overall survival as the primary outcome of interest, and adverse events as secondary outcome of interest were recorded for meta-analysis. RESULTS: A Bayesian network meta-analysis is an appropriate statistical method to compare all treatment options by statistically simulating the estimated results of a comprehensive trial, and to compare treatments by common and associated comparators. In addition, Bayesian network meta-analysis can produce ranking probabilities of treatments, which may contribute to clinicians' clinical decision-making.

Comparison of Essential Oils of Houttuynia cordata Thunb. from Different Processing Methods and Harvest Seasons Based on GC-MS and Chemometric Analysis.

Houttuyniae Herba (HH) refers to the dried aerial part of Houttuynia cordata Thunb. (DHC) or the fresh whole grass of Houttuynia cordata Thunb. (FHC), where DHC are harvested in summer and FHC around the year. However, harvest seasons and processing methods (i.e., medicinal parts and drying process) might affect the quality of HH. To compare the essential oils (EOs) of DHC and FHC and their two harvest seasons, GC-MS analysis combined with chemometric analysis was applied. The results showed that the oil yield of FHC (0.076 ± 0.030%) was higher than that of DHC (0.038 ± 0.029%), and oil yield was higher in summer than in autumn (0.044 ± 0.029% for DHC1, 0.036 ± 0.028% for DHC2, 0.084 ± 0.026% for FHC1, and 0.067 ± 0.033% for FHC2, respectively). Moreover, hierarchical cluster analysis (HCA) and principal component analysis (PCA) successfully distinguished the chemical constituents of DHC and FHC oils. Additionally, according to orthogonal partial least squares discriminant analysis (OPLS-DA), eleven components were selected as chemical markers for discriminating DHC and FHC, and two and four chemical markers for discriminating two harvest seasons of DHC and FHC, respectively. Among these markers, the average contents of α-pinene, limonene, ß-phellandrene, α-terpineol, 4-tridecanone, and ethyl decanoate were higher in FHC oils. In contrast, the average contents of nonanal, 1-nonanol, ß-cyclocitral, n-hexadecanoic acid, and octadecanol were higher in DHC oils. Additionally, the contents of 4-tridecanone and ethyl decanoate were both higher in DHC1 oils than in DHC2 oils. Moreover, the contents of ß-myrcene and ß-phellandrene were higher in FHC1 oils, while the contents of 2,6-octadien-1-ol, 3,7-dimethyl-, acetate, and (z)-phytol were higher in FHC2 oils. For these reasons, this study provides a scientific basis for quality control and clinical medication.

Neural-ODE for pharmacokinetics modeling and its advantage to alternative machine learning models in predicting new dosing regimens.

Forecasting pharmacokinetics (PK) for individual patients is a fundamental problem in clinical pharmacology. One key challenge is that PK models constructed using data from one dosing regimen must predict PK data for different dosing regimen(s). We propose a deep learning approach based on neural ordinary differential equations (neural-ODE) and tested its generalizability against a variety of alternative models. Specifically, we used the PK data from two different treatment regimens of trastuzumab emtansine. The models performed similarly when the training and the test sets come from the same dosing regimen. However, for predicting a new treatment regimen, the neural-ODE model showed substantially better performance. To date, neural-ODE is the most accurate PK model in predicting untested treatment regimens. This study represents the first time neural-ODE has been applied to PK modeling and the results suggest it is a widely applicable algorithm with the potential to impact future studies.

[Correlation between cold pain of knee joint and subchondral bone marrow edema in patients with knee osteoarthritis].

OBJECTIVE: To investigate relationship between cold pain of knee joint and subchondral bone marrow edema (BME). METHODS: From May 2018 to August 2019, 92 patients with knee osteoarthritis (KOA) associated with cold pain of knee were admitted, all patients were underwent MRI examination. The patients were divided into observation group (47 patients with BME) and control group(45 patients without BME). In observation group, there were 6 males and 41 females aged from 36 to 87 years old with an average of (63.2±12.3) years old. In control group, there were 10 males and 35 females, aged from 48 to 84 years old with an average of (62.7±8.3) years old. All patientswere treated with drugs. The degree of joint degeneration was evaluated by Kellgren-Lawrence (K-L) grading. Degree of cold pain of knee was evaluated by knee cold pain score, and degree of BME was evaluated according to WORMS. The correlation between cold pain of knee and K-L grading and BME was analyzed. RESULTS: Score of cold pain in observation group (15.55±7.68) was higher than that of control group (9.42± 5.50), which had significant difference (t=4.383, P<0.001). There was no correlation between cold pain of knee and K-L grading(χ2=2.138, P=0.907). There was correlation between BME grading and degree of cold pain in observation group(χ2=19.709, P<0.001), and Spearman correlation coefficient was rs=0.509(P<0.001). CONCLUSION: The cold pain of KOA patients is not related to K-L grading, but corelate with BME grading. The Cold pain of knee was more pronounced in KOA patients with BME, and the severity of BME is often related to degree of cold pain. It seemed to be a tendency:the more serious BME, the heavier coldpain.

Efficiency and mechanism of nitrogen removal from piggery wastewater in an improved microaerobic process.

Characterized by high ammonium (NH4+ - N) and low ratio of chemical oxygen demand (COD) to total nitrogen (COD/TN), discharge of piggery wastewater has been identified as a primary pollution source resulting in water eutrophication. An improved microaerobic reactor, internal aerating microaerobic reactor (IAMR), was constructed to treat manure-free piggery wastewater without effluent recycle at dissolved oxygen of 0.3 mg/L and 32 °C. A removal rate of COD, NH4+ - N and TN averaged 77.9%, 94.6% and 82.6% was obtained in the reactor, with the concentration of 258.5, 235.5 and 335.2 mg/L in influent, respectively. 16S rDNA amplicon sequencing, carbon and nitrogen mass balance and stoichiometry indicated that heterotrophic nitrification-anammox was the dominant approach to nitrogen removal. Microbiome phenotypes showed that aerobic bacteria were the dominant microorganisms, and the microbiome oxidative stress tolerance was intensified along with the continuous operation of the IAMR, resulting in the survival of various facultative and anaerobic bacteria for nutrients removal. With the good nutrients removal, less energy consumption, and high tolerance to influent fluctuation, the improved IAMR was confirmed as a promising process for treating wastewater with high NH4+ - N and low COD/TN.

RBM10 regulates human TERT gene splicing and inhibits pancreatic cancer progression.

Dysregulation of alternative splicing of hTERT gene to generate full-length Htert (hTERT-FL) that reactivate telomerase has been recognized as a major pathological alteration in pancreatic cancer (PrCa). Mechanism about the factors that regulate hTERT-FL splicing is lacking. Through bioinformatics approach, we focus on a candidate splicing factor RBM10, which leads to a switch in hTERT transcripts to generate a function-less isoform hTERT-s in PrCa, suppressed both telomerase activity and subsequent telomere shortening. RBM10 expression is negatively associated with PrCa progression. Gain or loss of RBM10 also significantly changed PrCa cell proliferation in vitro and in xenografts. RNA-IP and RNA pull-down assays reveal that RBM10 promotes the exclusion of exons7 and 8 which results in the production of TERT-s transcripts. This study may increase knowledge about potentially targetable cancer associated splicing factors and provide novel insights into therapeutic approach in PrCa.

Deep Learning Identifies Digital Biomarkers for Self-Reported Parkinson's Disease.

Large-scale population screening and in-home monitoring for patients with Parkinson's disease (PD) has so far been mainly carried out by traditional healthcare methods and systems. Development of mobile health may provide an independent, future method to detect PD. Current PD detection algorithms will benefit from better generalizability with data collected in real-world situations. In this paper, we report the top-performing smartphone-based method in the recent DREAM Parkinson's Disease Digital Biomarker Challenge for digital diagnosis of PD. Utilizing real-world accelerometer records, this approach differentiated PD from control subjects with an area under the receiver-operating characteristic curve of 0.87 by 3D augmentation of accelerometer records, a significant improvement over other state-of-the-art methods. This study paves the way for future at-home screening of PD and other neurodegenerative conditions affecting movement.

Treatment Stratification of Patients with Metastatic Castration-Resistant Prostate Cancer by Machine Learning.

Prostate cancer is the most common cancer in men in the Western world. One-third of the patients with prostate cancer will develop resistance to hormonal therapy and progress into metastatic castration-resistant prostate cancer (mCRPC). Currently, docetaxel is a preferred treatment for mCRPC. However, about 20% of the patients will undergo early therapeutic failure owing to adverse events induced by docetaxel-based chemotherapy. There is an emergent need for a computational model that can accurately stratify patients into docetaxel-tolerable and docetaxel-intolerable groups. Here we present the best-performing algorithm in the Prostate Cancer DREAM Challenge for predicting adverse events caused by docetaxel treatment. We integrated the survival status and severity of adverse events into our model, which is an innovative way to complement and stratify the treatment discontinuation information. Critical stratification biomarkers were further identified in determining the treatment discontinuation. Our model has the potential to improve future personalized treatment in mCRPC.