Independent Test Dataset Research Articles

Unified mRNA Subcellular Localization Predictor based on machine learning techniques

BackgroundThe mRNA subcellular localization bears substantial impact in the regulation of gene expression, cellular migration, and adaptation. However, the methods employed for experimental determination of this localization are arduous, time-intensive, and come with a high cost.MethodsIn this research article, we tackle the essential challenge of predicting the subcellular location of messenger RNAs (mRNAs) through Unified mRNA Subcellular Localization Predictor (UMSLP), a machine learning (ML) based approach. We embrace an in silico strategy that incorporate four distinct feature sets: kmer, pseudo k-tuple nucleotide composition, nucleotide physicochemical attributes, and the 3D sequence depiction achieved via Z-curve transformation for predicting subcellular localization in benchmark dataset across five distinct subcellular locales, encompassing nucleus, cytoplasm, extracellular region (ExR), mitochondria, and endoplasmic reticulum (ER).ResultsThe proposed ML model UMSLP attains cutting-edge outcomes in predicting mRNA subcellular localization. On independent testing dataset, UMSLP ahcieved over 87% precision, 94% specificity, and 94% accuracy. Compared to other existing tools, UMSLP outperformed mRNALocator, mRNALoc, and SubLocEP by 11%, 21%, and 32%, respectively on average prediction accuracy for all five locales. SHapley Additive exPlanations analysis highlights the dominance of k-mer features in predicting cytoplasm, nucleus, ER, and ExR localizations, while Z-curve based features play pivotal roles in mitochondria subcellular localization detection.AvailabilityWe have shared datasets, code, Docker API for users in GitHub at: https://github.com/smusleh/UMSLP.

Forest aboveground biomass estimation by GEDI and multi-source EO data fusion over Indian forest

ABSTRACT Monitoring changes in carbon stocks through forest biomass assessment is crucial for carbon cycle studies. However, challenges in obtaining timely and reliable ground measurements hinder creation of the spatially continuous maps of forest aboveground biomass density (AGBD). This study proposes an approach for generating spatially continuous maps of forest aboveground biomass density (AGBD) by combining Global Ecosystem Dynamics Investigation (GEDI) LiDAR-based data with open-access earth observation (EO) data. The key contribution of the study lies in the systematic evaluation of various model configurations to select the optimal model for AGBD generation. The evaluation considered various model configurations, including predictor sets, spatial resolution, beam selection, and sensitivity thresholds. We used a Random Forest model, trained through five-fold cross-validation on 80% of the total data, to estimate AGBD in the Indian forest region. Model performance was assessed using the 20% independent test dataset. Results, using Sentinel-1 and 2 predictors, yielded R2 values of 0.55 to 0.60 and RMSE of 48.5 to 56.3 Mg/ha. Incorporating forest and agroclimatic zone attributes improved performance (R2: 0.59 to 0.69, RMSE: 42.2 to 53.3 Mg/ha). The selection of the top 15 predictors, which favoured features from Sentinel-2, DEM, forest attributes, and agroclimatic zones, and GEDI data with sensitivity >0.98, yielded the optimal model with an R2 of 0.64 and RMSE of 46.59 Mg/ha. The results underscore the significance of incorporating attributes like forest and agro-climatic zones and the need for an optimal model selection considering predictor types and GEDI shot characteristics. The top-performing model is validated in Simdega, Jharkhand (R2: 0.74, RMSE: 39.3 Mg/ha), demonstrating the methodological potential of this approach. Overall, this study emphasizes the methodological prospects of integrating multi-source open-access EO data to produce spatially continuous aboveground biomass (AGB) maps through data fusion.

StructuralDPPIV: a novel deep learning model based on atom structure for predicting dipeptidyl peptidase-IV inhibitory peptides.

Diabetes is a chronic metabolic disorder that has been a major cause of blindness, kidney failure, heart attacks, stroke, and lower limb amputation across the world. To alleviate the impact of diabetes, researchers have developed the next generation of anti-diabetic drugs, known as dipeptidyl peptidase IV inhibitory peptides (DPP-IV-IPs). However, the discovery of these promising drugs has been restricted due to the lack of effective peptide-mining tools. Here, we presented StructuralDPPIV, a deep learning model designed for DPP-IV-IP identification, which takes advantage of both molecular graph features in amino acid and sequence information. Experimental results on the independent test dataset and two wet experiment datasets show that our model outperforms the other state-of-art methods. Moreover, to better study what StructuralDPPIV learns, we used CAM technology and perturbation experiment to analyze our model, which yielded interpretable insights into the reasoning behind prediction results. The project code is available at https://github.com/WeiLab-BioChem/Structural-DPP-IV.

Identifying virulence factors using graph transformer autoencoder with ESMFold-predicted structures

With the increasing resistance of bacterial pathogens to conventional antibiotics, antivirulence strategies targeting virulence factors (VFs) have become an effective new therapy for the treatment of pathogenic bacterial infections. Therefore, the identification and prediction of VFs can provide ideal candidate targets for the implementation of antivirulence strategies in treating infections caused by pathogenic bacteria. Currently, the existing computational models predominantly rely on the amino acid sequences of virulence proteins while overlooking structural information. Here, we propose a novel graph transformer autoencoder for VF identification (GTAE-VF), which utilizes ESMFold-predicted 3D structures and converts the VF identification problem into a graph-level prediction task. In an encoder-decoder framework, GTAE-VF adaptively learns both local and global information by integrating a graph convolutional network and a transformer to implement all-pair message passing, which can better capture long-range correlations and potential relationships. Extensive experiments on an independent test dataset demonstrate that GTAE-VF achieves reliable and robust prediction accuracy with an AUC of 0.963, which is consistently better than that of other structure-based and sequence-based approaches. We believe that GTAE-VF has the potential to emerge as a valuable tool for assessing VFs and devising antivirulence strategies.

DeepPRMS: advanced deep learning model to predict protein arginine methylation sites.

Protein methylation is a form of post-translational modifications of protein, which is crucial for various cellular processes, including transcription activity and DNA repair. Correctly predicting protein methylation sites is fundamental for research and drug discovery. Some experimental techniques, such as methyl-specific antibodies, chromatin immune precipitation and mass spectrometry, exist for predicting protein methylation sites, but these techniques are time-consuming and costly. The ability to predict methylation sites using in silico techniques may help researchers identify potential candidate sites for future examination and make it easier to carry out site-specific investigations and downstream characterizations. In this research, we proposed a novel deep learning-based predictor, named DeepPRMS, to identify protein methylation sites in primary sequences. The DeepPRMS utilizes the gated recurrent unit (GRU) and convolutional neural network (CNN) algorithms to extract the sequential and spatial information from the primary sequences. GRU is used to extract sequential information, while CNN is used for spatial information. We combined the latent representation of GRU and CNN models to have a better interaction among them. Based on the independent test data set, DeepPRMS obtained an accuracy of 85.32%, a specificity of 84.94%, Matthew's correlation coefficient of 0.71 and a sensitivity of 85.80%. The results indicate that DeepPRMS can predict protein methylation sites with high accuracy and outperform the state-of-the-art models. The DeepPRMS is expected to effectively guide future research experiments for identifying potential methylated protein sites. The web server is available at http://deepprms.nitsri.ac.in/.

Crescent calculator: A webtool enabling objective decision-making for assessment of IgA nephropathy immune activity throughout the disease course

IgA nephropathy (IgAN) is an immune-mediated glomerulonephritis, posing a challenge for the long-term management. It is crucial to monitor the disease’s activity over the disease course. Crescent lesions have been known as an active lesion associated with immune activity. We aimed to develop the Crescent Calculator to aid clinicians in making timely and well-informed decisions throughout the long-term disease course, such as renal biopsies and immunosuppressive therapy.1,761 patients with biopsy-proven IgAN were recruited from four medical centers in Zhejiang Province, China. 16.9% presented crescent lesions. UPCR, URBC, eGFR and C4 were independently associated with the crescent lesions. By incorporating these variables, the Crescent Calculator was constructed to estimate the likelihood of crescent lesions. The predictor achieved AUC values of over 0.82 in two independent testing datasets. In addition, to fulfill varied clinical needs, multiple classification modes were established.The Crescent Calculator was developed to estimate the risk of crescent lesions for patients with IgAN, assisting clinicians in making timely, objective, and well-informed decisions regarding the need for renal biopsies and more appropriate use of immunosuppressive therapy in patients with IgAN.

A clinical consensus-compliant deep learning approach to quantitatively evaluate human in vitro fertilization early embryonic development with optical microscope images

The selection of embryos is a key for the success of in vitro fertilization (IVF). However, automatic quality assessment on human IVF embryos with optical microscope images is still challenging. In this study, we developed a clinical consensus-compliant deep learning approach, named Esava (Embryo Segmentation and Viability Assessment), to quantitatively evaluate the development of IVF embryos using optical microscope images. In total 551 optical microscope images of human IVF embryos of day-2 to day-3 were collected, preprocessed, and annotated. Using the Faster R-CNN model as baseline, our Esava model was constructed, refined, trained, and validated for precise and robust blastomere detection. A novel algorithm Crowd-NMS was proposed and employed in Esava to enhance the object detection and to precisely quantify the embryonic cells and their size uniformity. Additionally, an innovative GrabCut-based unsupervised module was integrated for the segmentation of blastomeres and embryos. Independently tested on 94 embryo images for blastomere detection, Esava obtained the high rates of 0.9940, 0.9121, and 0.9531 for precision, recall, and mAP respectively, and gained significant advances compared with previous computational methods. Intraclass correlation coefficients indicated the consistency between Esava and three experienced embryologists. Another test on 51 extra images demonstrated that Esava surpassed other tools significantly, achieving the highest average precision 0.9025. Moreover, it also accurately identified the borders of blastomeres with mIoU over 0.88 on the independent testing dataset. Esava is compliant with the Istanbul clinical consensus and compatible to senior embryologists. Taken together, Esava improves the accuracy and efficiency of embryonic development assessment with optical microscope images.

Multitask learning for predicting pulmonary absorption of chemicals

Pulmonary absorption is an important route for drug delivery and chemical exposure. To streamline the chemical assessment process for the reduction of animal experiments, several animal-free models were developed for pulmonary absorption research. While Calu-3 and Caco-2 cells and their derived computational models were used in estimating pulmonary permeability, the ex vivo isolated perfused lung (IPL) models are considered more clinically relevant measurements. However, the IPL experiments are resource-consuming making it infeasible for the large-scale screening of potential inhaled toxicants and drugs. In silico models are desirable for estimating pulmonary absorption. This study presented a novel machine learning method that employed an extratrees-based multitask learning approach to predict the IPL absorption rate constant (kaIPL) of various chemicals. The shared permeability knowledge was extracted by simultaneously learning three relevant tasks of Caco-2 and Calu-3 cell permeability and IPL absorption rate. Seven informative physicochemical descriptors were identified. A rigorous evaluation of the developed prediction model showed good performance with a high correlation between predictions and observations (r = 0.84) in the independent test dataset. Two case studies of inhalation drugs and respiratory sensitizers revealed the potential application of this model, which may serve as a valuable tool for predicting pulmonary absorption of chemicals.

AGF-PPIS: A protein–protein interaction site predictor based on an attention mechanism and graph convolutional networks

Protein–protein interactions play an important role in various biological processes. Interaction among proteins has a wide range of applications. Therefore, the correct identification of protein–protein interactions sites is crucial. In this paper, we propose a novel predictor for protein–protein interactions sites, AGF-PPIS, where we utilize a multi-head self-attention mechanism (introducing a graph structure), graph convolutional network, and feed-forward neural network. We use the Euclidean distance between each protein residue to generate the corresponding protein graph as the input of AGF-PPIS. On the independent test dataset Test_60, AGF-PPIS achieves superior performance over comparative methods in terms of seven different evaluation metrics (ACC, precision, recall, F1-score, MCC, AUROC, AUPRC), which fully demonstrates the validity and superiority of the proposed AGF-PPIS model. The source codes and the steps for usage of AGF-PPIS are available at https://github.com/fxh1001/AGF-PPIS.

Visualization of Optic Nerve Structural Patterns in Papilledema Using Deep Learning Variational Autoencoders.

To visualize and quantify structural patterns of optic nerve edema encountered in papilledema during treatment. A novel bi-channel deep-learning variational autoencoder (biVAE) model was trained using 1498 optical coherence tomography (OCT) scans of 125 subjects over time from the Idiopathic Intracranial Hypertension Treatment Trial (IIHTT) and 791 OCT scans of 96 control subjects from the University of Iowa. An independent test dataset of 70 eyes from 70 papilledema subjects was used to evaluate the ability of the biVAE model to quantify and reconstruct the papilledema spatial patterns from input OCT scans using only two variables. The montage color maps of the retinal nerve fiber layer (RNFL) and total retinal thickness (TRT) produced by the biVAE model provided an organized visualization of the variety of morphological patterns of optic disc edema (including differing patterns at similar thickness levels). Treatment effects of acetazolamide versus placebo in the IIHTT were also demonstrated in the latent space. In image reconstruction, the mean signed peripapillary retinal nerve fiber layer thickness (pRNFLT) difference ± SD was -0.12 ± 17.34 µm, the absolute pRNFLT difference was 13.68 ± 10.65 µm, and the RNFL structural similarity index reached 0.91 ± 0.05. A wide array of structural patterns of papilledema, integrating the magnitude of disc edema with underlying disc and retinal morphology, can be quantified by just two latent variables. A biVAE model encodes structural patterns, as well as the correlation between channels, and may be applied to visualize individuals or populations with papilledema throughout treatment.

Machine learning and feature extraction for rapid antimicrobial resistance prediction of Acinetobacter baumannii from whole-genome sequencing data.

Whole-genome sequencing (WGS) has contributed significantly to advancements in machine learning methods for predicting antimicrobial resistance (AMR). However, the comparisons of different methods for AMR prediction without requiring prior knowledge of resistance remains to be conducted. We aimed to predict the minimum inhibitory concentrations (MICs) of 13 antimicrobial agents against Acinetobacter baumannii using three machine learning algorithms (random forest, support vector machine, and XGBoost) combined with k-mer features extracted from WGS data. A cohort of 339 isolates was used for model construction. The average essential agreement and category agreement of the best models exceeded 90.90% (95%CI, 89.03-92.77%) and 95.29% (95%CI, 94.91-95.67%), respectively; the exceptions being levofloxacin, minocycline and imipenem. The very major error rates ranged from 0.0 to 5.71%. We applied feature selection pipelines to extract the top-ranked 11-mers to optimise training time and computing resources. This approach slightly improved the prediction performance and enabled us to obtain prediction results within 10 min. Notably, when employing these top-ranked 11-mers in an independent test dataset (120 isolates), we achieved an average accuracy of 0.96. Our study is the first to demonstrate that AMR prediction for A. baumannii using machine learning methods based on k-mer features has competitive performance over traditional workflows; hence, sequence-based AMR prediction and its application could be further promoted. The k-mer-based workflow developed in this study demonstrated high recall/sensitivity and specificity, making it a dependable tool for MIC prediction in clinical settings.

Effective strategies for correcting spatial sampling bias in species distribution models without independent test data

AbstractAimSpatial sampling bias (SSB) is a feature of opportunistically sampled species records. Species distribution models (SDMs) built using these data (i.e. presence‐background models) can produce biased predictions of suitability across geographic space, confounding species occurrence with the distribution of sampling effort. A wide range of SSB correction methods have been developed but simulations suggest effects on predictive performance are highly variable. Here, we aim to identify the SSB correction methods that have the highest likelihood of improving true predictive performance and evaluation strategies that provide a reliable indicator of model performance when independent test data are unavailable.LocationGlobal, simulation.Time PeriodCurrent, simulation.MethodsA meta‐analysis was used to evaluate the performance of SSB correction methods in studies where there were direct comparisons between corrected and uncorrected SDMs. A simulation model was then developed to test evaluation strategies against a known truth using four common SSB correction methods.ResultsEffect sizes from published studies suggest some support for small positive effects of SSB correction on predictive performance when assessed using independent test data, but this was not evident using internal cross‐validation and no single method stood out as consistently effective. Simulations support these findings and show that evaluation using internal test data was generally a poor indicator of the true effect of SSB correction. Methods that adjust models relative to a known driver of SSB produced the largest performance gains, but were also the most inconsistent.Main ConclusionsCorrecting SSB in presence‐background SDMs without independent test data to evaluate the effect on model performance requires careful implementation. We recommend clearer documentation of SSB correction effects on SDMs, presenting results from models with and without correction, evaluating effects of different assumptions of SSB implementation on predictions, as well as greater efforts to collect independent test datasets to validate model predictions.

ISUMO-RsFPN: A predictor for identifying lysine SUMOylation sites based on multi-features and feature pyramid networks

SUMOylation is a protein post-translational modification that plays an essential role in cellular functions. For predicting SUMO sites, numerous researchers have proposed advanced methods based on ordinary machine learning algorithms. These reported methods have shown excellent predictive performance, but there is room for improvement. In this study, we constructed a novel deep neural network Residual Pyramid Network (RsFPN), and developed an ensemble deep learning predictor called iSUMO-RsFPN. Initially, three feature extraction methods were employed to extract features from samples. Following this, weak classifiers were trained based on RsFPN for each feature type. Ultimately, the weak classifiers were integrated to construct the final classifier. Moreover, the predictor underwent systematically testing on an independent test dataset, where the results demonstrated a significant improvement over the existing state-of-the-art predictors. The code of iSUMO-RsFPN is free and available at https://github.com/454170054/iSUMO-RsFPN.

SumoPred-PLM: human SUMOylation and SUMO2/3 sites Prediction using Pre-trained Protein Language Model.

SUMOylation is an essential post-translational modification system with the ability to regulate nearly all aspects of cellular physiology. Three major paralogues SUMO1, SUMO2and SUMO3 form a covalent bond between the small ubiquitin-like modifier with lysine residues at consensus sites in protein substrates. Biochemical studies continue to identify unique biological functions for protein targets conjugated to SUMO1 versus the highly homologous SUMO2 and SUMO3 paralogues. Yet, the field has failed to harness contemporary AI approaches including pre-trained protein language models to fully expand and/or recognize the SUMOylated proteome. Herein, we present a novel, deep learning-based approach called SumoPred-PLM for human SUMOylation prediction with sensitivity, specificity, Matthew's correlation coefficient, and accuracy of 74.64%, 73.36%, 0.48% and 74.00%, respectively, on the CPLM 4.0 independent test dataset. In addition, this novel platform uses contextualized embeddings obtained from a pre-trained protein language model, ProtT5-XL-UniRef50 to identify SUMO2/3-specific conjugation sites. The results demonstrate that SumoPred-PLM is a powerful and unique computational tool to predict SUMOylation sites in proteins and accelerate discovery.

STM-ac4C: a hybrid model for identification of N4-acetylcytidine (ac4C) in human mRNA based on selective kernel convolution, temporal convolutional network, and multi-head self-attention.

N4-acetylcysteine (ac4C) is a chemical modification in mRNAs that alters the structure and function of mRNA by adding an acetyl group to the N4 position of cytosine. Researchers have shown that ac4C is closely associated with the occurrence and development of various cancers. Therefore, accurate prediction of ac4C modification sites on human mRNA is crucial for revealing its role in diseases and developing new diagnostic and therapeutic strategies. However, existing deep learning models still have limitations in prediction accuracy and generalization ability, which restrict their effectiveness in handling complex biological sequence data. This paper introduces a deep learning-based model, STM-ac4C, for predicting ac4C modification sites on human mRNA. The model combines the advantages of selective kernel convolution, temporal convolutional networks, and multi-head self-attention mechanisms to effectively extract and integrate multi-level features of RNA sequences, thereby achieving high-precision prediction of ac4C sites. On the independent test dataset, STM-ac4C showed improvements of 1.81%, 3.5%, and 0.37% in accuracy, Matthews correlation coefficient, and area under the curve, respectively, compared to the existing state-of-the-art technologies. Moreover, its performance on additional balanced and imbalanced datasets also confirmed the model's robustness and generalization ability. Various experimental results indicate that STM-ac4C outperforms existing methods in predictive performance. In summary, STM-ac4C excels in predicting ac4C modification sites on human mRNA, providing a powerful new tool for a deeper understanding of the biological significance of mRNA modifications and cancer treatment. Additionally, the model reveals key sequence features that influence the prediction of ac4C sites through sequence region impact analysis, offering new perspectives for future research. The source code and experimental data are available at https://github.com/ymy12341/STM-ac4C.

An individualized stemness-related signature to predict prognosis and immunotherapy responses for gastric cancer using single-cell and bulk tissue transcriptomes.

Currently, many stemness-related signatures have been developed for gastric cancer (GC) to predict prognosis and immunotherapy outcomes. However, due to batch effects, these signatures cannot accurately analyze patients one by one, rendering them impractical in real clinical scenarios. Therefore, we aimed to develop an individualized and clinically applicable signature based on GC stemness. Malignant epithelial cells from single-cell RNA-Seq data of GC were used to identify stemness-related signature genes based on the CytoTRACE score. Using two bulk tissue datasets as training data, the enrichment scores of the signature genes were applied to classify samples into two subtypes. Then, using the identified subtypes as criteria, we developed an individualized stemness-related signature based on the within-sample relative expression orderings of genes. We identified 175 stemness-related signature genes, which exhibited significantly higher AUCell scores in poorly differentiated GCs compared to differentiated GCs. In training datasets, GC samples were classified into two subtypes with significantly different survival times and genomic characteristics. Utilizing the two subtypes, an individualized signature was constructed containing 47 gene pairs. In four independent testing datasets, GC samples classified as high risk exhibited significantly shorter survival times, higher infiltration of M2 macrophages, and lower immune responses compared to low-risk samples. Moreover, the potential therapeutic targets and corresponding drugs were identified for the high-risk group, such as CD248 targeted by ontuxizumab. We developed an individualized stemness-related signature, which can accurately predict the prognosis and efficacy of immunotherapy for each GC sample.

A Physics-informed Deep Neural Network for Harmonization of CT Images.

An adversarial generative network was trained on virtual CT images acquired under various imaging conditions using a virtual imaging platform with 40 computational patient models. These models featured anthropomorphic lungs with different levels of pulmonary diseases, including nodules and emphysema. Imaging was conducted using a validated CT simulator at two dose levels and varying reconstruction kernels. The trained model was tested on an independent virtual test dataset and two clinical datasets. On the virtual test set, the harmonizer improved the structural similarity index from 79.3 ±16.4% to 95.8 ±1.7%, normalized mean squared error from 16.7 ±9.7% to 9.2 ±1.7%, and peak signal-to-noise ratio from 27.7 ±3.7dB to 32.2 ±1.6dB. Moreover, the harmonized images yielded more precise quantification of emphysema-based imaging biomarkers for lung attenuation, LAA-950 from 5.6 ±8.7% to 0.23 ±0.16%, Perc15 from 43.4 ±45.4 HU to 20.0 ±7.5 HU, and Lung Mass from 0.3 ±0.3 g to 0.1 ±0.2 g. In clinical data, the harmonizer reduced biomarker variability by an average of 70%. For lung nodules, harmonized images improved the detectability index by 6.5-fold and DNN-based precision by 6%. The proposed harmonizer significantly enhances image quality and quantification accuracy in CT imaging. The study demonstrated the potential utility of image harmonization for consistent CT image quality and reliable quantification, which is crucial for clinical applications and patient management.

External evaluation of the Dynamic Criticality Index: A machine learning model to predict future need for ICU care in hospitalized pediatric patients.

To assess the single site performance of the Dynamic Criticality Index (CI-D) models developed from a multi-institutional database to predict future care. Secondarily, to assess future care-location predictions in a single institution when CI-D models are re-developed using single-site data with identical variables and modeling methods. Four CI-D models were assessed for predicting care locations >6-12 hours, >12-18 hours, >18-24 hours, and >24-30 hours in the future. Prognostic study comparing multi-institutional CI-D models' performance in a single-site electronic health record dataset to an institution-specific CI-D model developed using identical variables and modelling methods. The institution did not participate in the multi-institutional dataset. All pediatric inpatients admitted from January 1st 2018 -February 29th 2020 through the emergency department. The main outcome was inpatient care in routine or ICU care locations. A total of 29,037 pediatric hospital admissions were included, with 5,563 (19.2%) admitted directly to the ICU, 869 (3.0%) transferred from routine to ICU care, and 5,023 (17.3%) transferred from ICU to routine care. Patients had a median [IQR] age 68 months (15-157), 47.5% were female and 43.4% were black. The area under the receiver operating characteristic curve (AUROC) for the multi-institutional CI-D models applied to a single-site test dataset was 0.493-0.545 and area under the precision-recall curve (AUPRC) was 0.262-0.299. The single-site CI-D models applied to an independent single-site test dataset had an AUROC 0.906-0.944 and AUPRC range from 0.754-0.824. Accuracy at 0.95 sensitivity for those transferred from routine to ICU care was 72.6%-81.0%. Accuracy at 0.95 specificity was 58.2%-76.4% for patients who transferred from ICU to routine care. Models developed from multi-institutional datasets and intended for application to individual institutions should be assessed locally and may benefit from re-development with site-specific data prior to deployment.

DeepAMPNet: a novel antimicrobial peptide predictor employing AlphaFold2 predicted structures and a bi-directional long short-term memory protein language model.

Global public health is seriously threatened by the escalating issue of antimicrobial resistance (AMR). Antimicrobial peptides (AMPs), pivotal components of the innate immune system, have emerged as a potent solution to AMR due to their therapeutic potential. Employing computational methodologies for the prompt recognition of these antimicrobial peptides indeed unlocks fresh perspectives, thereby potentially revolutionizing antimicrobial drug development. In this study, we have developed a model named as deepAMPNet. This model, which leverages graph neural networks, excels at the swift identification of AMPs. It employs structures of antimicrobial peptides predicted by AlphaFold2, encodes residue-level features through a bi-directional long short-term memory (Bi-LSTM) protein language model, and constructs adjacency matrices anchored on amino acids' contact maps. In a comparative study with other state-of-the-art AMP predictors on two external independent test datasets, deepAMPNet outperformed in accuracy. Furthermore, in terms of commonly accepted evaluation matrices such as AUC, Mcc, sensitivity, and specificity, deepAMPNet achieved the highest or highly comparable performances against other predictors. deepAMPNet interweaves both structural and sequence information of AMPs, stands as a high-performance identification model that propels the evolution and design in antimicrobial peptide pharmaceuticals. The data and code utilized in this study can be accessed at https://github.com/Iseeu233/deepAMPNet.

DMAMP: A deep-learning model for detecting antimicrobial peptides and their multi-activities.

Due to the broad-spectrum and high-efficiency antibacterial activity, antimicrobial peptides (AMPs) and their functions have been studied in the field of drug discovery. Using biological experiments to detect the AMPs and corresponding activities require a high cost, whereas computational technologies do so for much less. Currently, most computational methods solve the identification of AMPs and their activities as two independent tasks, which ignore the relationship between them. Therefore, the combination and sharing of patterns for two tasks is a crucial problem that needs to be addressed. In this study, we propose a deep learning model, called DMAMP, for detecting AMPs and activities simultaneously, which is benefited from multi-task learning. The first stage is to utilize convolutional neural network models and residual blocks to extract the sharing hidden features from two related tasks. The next stage is to use two fully connected layers to learn the distinct information of two tasks. Meanwhile, the original evolutionary features from the peptide sequence are also fed to the predictor of the second task to complement the forgotten information. The experiments on the independent test dataset demonstrate that our method performs better than the single-task model with 4.28% of Matthews Correlation Coefficient (MCC) on the first task, and achieves 0.2627 of an average MCC which is higher than the single-task model and two existing methods for five activities on the second task. To understand whether features derived from the convolutional layers of models capture the differences between target classes, we visualize these high-dimensional features by projecting into 3D space. In addition, we show that our predictor has the ability to identify peptides that achieve activity against Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2). We hope that our proposed method can give new insights into the discovery of novel antiviral peptide drugs.