Integrating genome-scale metabolic modelling and transfer learning for human gene regulatory network reconstruction

Emerging methods for genome-scale metabolic modeling of microbial communities

Genome-scale metabolic models (GEMs) are consolidating as platforms for studying mixed microbial populations, by combining biological data and knowledge with mathematical rigor. However, deploying these models to answer research questions can be challenging due to the increasing number of available computational tools, the lack of universal standards, and their inherent limitations. Here, we present a comprehensive overview of foundational concepts for building and evaluating genome-scale models of microbial communities. We then compare tools in terms of requirements, capabilities, and applications. Next, we highlight the current pitfalls and open challenges to consider when adopting existing tools and developing new ones. Our compendium can be relevant for the expanding community of modelers, both at the entry and experienced levels.

Integration of graph neural networks and genome-scale metabolic models for predicting gene essentiality

Genome-scale metabolic models are powerful tools for understanding cellular physiology. Flux balance analysis (FBA), in particular, is an optimization-based approach widely employed for predicting metabolic phenotypes. In model microbes such as Escherichia coli, FBA has been successful at predicting essential genes, i.e. those genes that impair survival when deleted. A central assumption in this approach is that both wild type and deletion strains optimize the same fitness objective. Although the optimality assumption may hold for the wild type metabolic network, deletion strains are not subject to the same evolutionary pressures and knock-out mutants may steer their metabolism to meet other objectives for survival. Here, we present FlowGAT, a hybrid FBA-machine learning strategy for predicting essentiality directly from wild type metabolic phenotypes. The approach is based on graph-structured representation of metabolic fluxes predicted by FBA, where nodes correspond to enzymatic reactions and edges quantify the propagation of metabolite mass flow between a reaction and its neighbours. We integrate this information into a graph neural network that can be trained on knock-out fitness assay data. Comparisons across different model architectures reveal that FlowGAT predictions for E. coli are close to those of FBA for several growth conditions. This suggests that essentiality of enzymatic genes can be predicted by exploiting the inherent network structure of metabolism. Our approach demonstrates the benefits of combining the mechanistic insights afforded by genome-scale models with the ability of deep learning to infer patterns from complex datasets.

COSMIC-dFBA: A novel multi-scale hybrid framework for bioprocess modeling

Metabolism governs cell performance in biomanufacturing, as it fuels growth and productivity. However, even in well-controlled culture systems, metabolism is dynamic, with shifting objectives and resources, thus limiting the predictive capability of mechanistic models for process design and optimization. Here, we present Cellular Objectives and State Modulation In bioreaCtors (COSMIC)-dFBA, a hybrid multi-scale modeling paradigm that accurately predicts cell density, antibody titer, and bioreactor metabolite concentration profiles. Using machine-learning, COSMIC-dFBA decomposes the instantaneous metabolite uptake and secretion rates in a bioreactor into weighted contributions from each cell state (growth or antibody-producing state) and integrates these with a genome-scale metabolic model. A major strength of COSMIC-dFBA is that it can be parameterized with only metabolite concentrations from spent media, although constraining the metabolic model with other omics data can further improve its capabilities. Using COSMIC-dFBA, we can predict the final cell density and antibody titer to within 10% of the measured data, and compared to a standard dFBA model, we found the framework showed a 90% and 72% improvement in cell density and antibody titer prediction, respectively. Thus, we demonstrate our hybrid modeling framework effectively captures cellular metabolism and expands the applicability of dFBA to model the dynamic conditions in a bioreactor.

Automatically transferring supervised targets method for segmenting lung lesion regions with CT imaging

BACKGROUND

To present an approach that autonomously identifies and selects a self-selective optimal target for the purpose of enhancing learning efficiency to segment infected regions of the lung from chest computed tomography images. We designed a semi-supervised dual-branch framework for training, where the training set consisted of limited expert-annotated data and a large amount of coarsely annotated data that was automatically segmented based on Hu values, which were used to train both strong and weak branches. In addition, we employed the Lovasz scoring method to automatically switch the supervision target in the weak branch and select the optimal target as the supervision object for training. This method can use noisy labels for rapid localization during the early stages of training, and gradually use more accurate targets for supervised training as the training progresses. This approach can utilize a large number of samples that do not require manual annotation, and with the iterations of training, the supervised targets containing noise become closer and closer to the fine-annotated data, which significantly improves the accuracy of the final model.

RESULTS

The proposed dual-branch deep learning network based on semi-supervision together with cost-effective samples achieved 83.56 ± 12.10 and 82.67 ± 8.04 on our internal and external test benchmarks measured by the mean Dice similarity coefficient (DSC). Through experimental comparison, the DSC value of the proposed algorithm was improved by 13.54% and 2.02% on the internal benchmark and 13.37% and 2.13% on the external benchmark compared with U-Net without extra sample assistance and the mean-teacher frontier algorithm, respectively.

CONCLUSION

The cost-effective pseudolabeled samples assisted the training of DL models and achieved much better results compared with traditional DL models with manually labeled samples only. Furthermore, our method also achieved the best performance compared with other up-to-date dual branch structures.

Machine Learning Methods for Small Data Challenges in Molecular Science

Small data are often used in scientific and engineering research due to the presence of various constraints, such as time, cost, ethics, privacy, security, and technical limitations in data acquisition. However, big data have been the focus for the past decade, small data and their challenges have received little attention, even though they are technically more severe in machine learning (ML) and deep learning (DL) studies. Overall, the small data challenge is often compounded by issues, such as data diversity, imputation, noise, imbalance, and high-dimensionality. Fortunately, the current big data era is characterized by technological breakthroughs in ML, DL, and artificial intelligence (AI), which enable data-driven scientific discovery, and many advanced ML and DL technologies developed for big data have inadvertently provided solutions for small data problems. As a result, significant progress has been made in ML and DL for small data challenges in the past decade. In this review, we summarize and analyze several emerging potential solutions to small data challenges in molecular science, including chemical and biological sciences. We review both basic machine learning algorithms, such as linear regression, logistic regression (LR), k-nearest neighbor (KNN), support vector machine (SVM), kernel learning (KL), random forest (RF), and gradient boosting trees (GBT), and more advanced techniques, including artificial neural network (ANN), convolutional neural network (CNN), U-Net, graph neural network (GNN), Generative Adversarial Network (GAN), long short-term memory (LSTM), autoencoder, transformer, transfer learning, active learning, graph-based semi-supervised learning, combining deep learning with traditional machine learning, and physical model-based data augmentation. We also briefly discuss the latest advances in these methods. Finally, we conclude the survey with a discussion of promising trends in small data challenges in molecular science.

Forecasting and what-if analysis of new positive COVID-19 cases during the first three waves in Italy

The joint exploitation of data related to epidemiological, mobility, and restriction aspects of COVID-19 with machine learning algorithms can support the development of predictive models that can be used to forecast new positive cases and study the impact of more or less severe restrictions. In this work, we integrate heterogeneous data from several sources and solve a multivariate time series forecasting task, specifically targeting the Italian case at both national and regional levels, during the first three waves of the pandemic. The goal is to build a robust predictive model to predict the number of new cases over a given time horizon so that any restrictive actions can be better planned. In addition, we perform a what-if analysis based on the best-identified predictive models to evaluate the impact of specific restrictions on the trend of positive cases. Our focus on the first three waves is motivated by the fact that it represents a typical emergency scenario (when no stable cure or vaccine is available) that may occur when a new pandemic spreads. Our experimental results prove that exploiting the considered heterogeneous data leads to accurate predictive models, reaching a WAPE of 5.75% at the national level. Furthermore, in the subsequent what-if analysis, we observed that strong all-in-one initiatives, such as total lockdowns, may not be adequate, while more specific and targeted solutions should be adopted. The developed models can help policy and decision-makers better plan intervention strategies and retrospectively analyze the effects of the decisions made at different scales. Joint exploitation of data on epidemiological, mobility, and restriction aspects of COVID-19 with machine learning algorithms to learn predictive models to forecast new positive cases.

Multi-dimensional experimental and computational exploration of metabolism pinpoints complex probiotic interactions

Multi-strain probiotics are widely regarded as effective products for improving gut microbiota stability and host health, providing advantages over single-strain probiotics. However, in general, it is unclear to what extent different strains would cooperate or compete for resources, and how the establishment of a common biofilm microenvironment could influence their interactions. In this work, we develop an integrative experimental and computational approach to comprehensively assess the metabolic functionality and interactions of probiotics across growth conditions. Our approach combines co-culture assays with genome-scale modelling of metabolism and multivariate data analysis, thus exploiting complementary data- and knowledge-driven systems biology techniques. To show the advantages of the proposed approach, we apply it to the study of the interactions between two widely used probiotic strains of Lactobacillus reuteri and Saccharomyces boulardii, characterising their production potential for compounds that can be beneficial to human health. Our results show that these strains can establish a mixed cooperative-antagonistic interaction best explained by competition for shared resources, with an increased individual exchange but an often decreased net production of amino acids and short-chain fatty acids. Overall, our work provides a strategy that can be used to explore microbial metabolic fingerprints of biotechnological interest, capable of capturing multifaceted equilibria even in simple microbial consortia.

Using transfer learning and dimensionality reduction techniques to improve generalisability of machine-learning predictions of mosquito ages from mid-infrared spectra

BACKGROUND

Old mosquitoes are more likely to transmit malaria than young ones. Therefore, accurate prediction of mosquito population age can drastically improve the evaluation of mosquito-targeted interventions. However, standard methods for age-grading mosquitoes are laborious and costly. We have shown that Mid-infrared spectroscopy (MIRS) can be used to detect age-specific patterns in mosquito cuticles and thus can be used to train age-grading machine learning models. However, these models tend to transfer poorly across populations. Here, we investigate whether applying dimensionality reduction and transfer learning to MIRS data can improve the transferability of MIRS-based predictions for mosquito ages.

METHODS

We reared adults of the malaria vector Anopheles arabiensis in two insectaries. The heads and thoraces of female mosquitoes were scanned using an attenuated total reflection-Fourier transform infrared spectrometer, which were grouped into two different age classes. The dimensionality of the spectra data was reduced using unsupervised principal component analysis or t-distributed stochastic neighbour embedding, and then used to train deep learning and standard machine learning classifiers. Transfer learning was also evaluated to improve transferability of the models when predicting mosquito age classes from new populations.

RESULTS

Model accuracies for predicting the age of mosquitoes from the same population as the training samples reached 99% for deep learning and 92% for standard machine learning. However, these models did not generalise to a different population, achieving only 46% and 48% accuracy for deep learning and standard machine learning, respectively. Dimensionality reduction did not improve model generalizability but reduced computational time. Transfer learning by updating pre-trained models with 2% of mosquitoes from the alternate population improved performance to ~ 98% accuracy for predicting mosquito age classes in the alternative population.

CONCLUSION

Combining dimensionality reduction and transfer learning can reduce computational costs and improve the transferability of both deep learning and standard machine learning models for predicting the age of mosquitoes. Future studies should investigate the optimal quantities and diversity of training data necessary for transfer learning and the implications for broader generalisability to unseen datasets.

Predicting drug characteristics using biomedical text embedding

BACKGROUND

Drug-drug interactions (DDIs) are preventable causes of medical injuries and often result in doctor and emergency room visits. Previous research demonstrates the effectiveness of using matrix completion approaches based on known drug interactions to predict unknown Drug-drug interactions. However, in the case of a new drug, where there is limited or no knowledge regarding the drug's existing interactions, such an approach is unsuitable, and other drug's preferences can be used to accurately predict new Drug-drug interactions.

METHODS

We propose adjacency biomedical text embedding (ABTE) to address this limitation by using a hybrid approach which combines known drugs' interactions and the drug's biomedical text embeddings to predict the DDIs of both new and well known drugs.

RESULTS

Our evaluation demonstrates the superiority of this approach compared to recently published DDI prediction models and matrix factorization-based approaches. Furthermore, we compared the use of different text embedding methods in ABTE, and found that the concept embedding approach, which involves biomedical information in the embedding process, provides the highest performance for this task. Additionally, we demonstrate the effectiveness of leveraging biomedical text embedding for additional drugs' biomedical prediction task by presenting text embedding's contribution to a multi-modal pregnancy drug safety classification.

CONCLUSION

Text and concept embeddings created by analyzing a domain-specific large-scale biomedical corpora can be used for predicting drug-related properties such as Drug-drug interactions and drug safety prediction. Prediction models based on the embeddings resulted in comparable results to hand-crafted features, however text embeddings do not require manual categorization or data collection and rely solely on the published literature.

Clinical stratification improves the diagnostic accuracy of small omics datasets within machine learning and genome-scale metabolic modelling methods

BACKGROUND

Recently, multi-omic machine learning architectures have been proposed for the early detection of cancer. However, for rare cancers and their associated small datasets, it is still unclear how to use the available multi-omics data to achieve a mechanistic prediction of cancer onset and progression, due to the limited data available. Hepatoblastoma is the most frequent liver cancer in infancy and childhood, and whose incidence has been lately increasing in several developed countries. Even though some studies have been conducted to understand the causes of its onset and discover potential biomarkers, the role of metabolic rewiring has not been investigated in depth so far.

METHODS

Here, we propose and implement an interpretable multi-omics pipeline that combines mechanistic knowledge from genome-scale metabolic models with machine learning algorithms, and we use it to characterise the underlying mechanisms controlling hepatoblastoma.

RESULTS AND CONCLUSIONS

While the obtained machine learning models generally present a high diagnostic classification accuracy, our results show that the type of omics combinations used as input to the machine learning models strongly affects the detection of important genes, reactions and metabolic pathways linked to hepatoblastoma. Our method also suggests that, in the context of computer-aided diagnosis of cancer, optimal diagnostic accuracy can be achieved by adopting a combination of omics that depends on the patient's clinical characteristics.

Transfer learning for genotype-phenotype prediction using deep learning models

BACKGROUND

For some understudied populations, genotype data is minimal for genotype-phenotype prediction. However, we can use the data of some other large populations to learn about the disease-causing SNPs and use that knowledge for the genotype-phenotype prediction of small populations. This manuscript illustrated that transfer learning is applicable for genotype data and genotype-phenotype prediction.

RESULTS

Using HAPGEN2 and PhenotypeSimulator, we generated eight phenotypes for 500 cases/500 controls (CEU, large population) and 100 cases/100 controls (YRI, small populations). We considered 5 (4 phenotypes) and 10 (4 phenotypes) different risk SNPs for each phenotype to evaluate the proposed method. The improved accuracy with transfer learning for eight different phenotypes was between 2 and 14.2 percent. The two-tailed p-value between the classification accuracies for all phenotypes without transfer learning and with transfer learning was 0.0306 for five risk SNPs phenotypes and 0.0478 for ten risk SNPs phenotypes.

CONCLUSION

The proposed pipeline is used to transfer knowledge for the case/control classification of the small population. In addition, we argue that this method can also be used in the realm of endangered species and personalized medicine. If the large population data is extensive compared to small population data, expect transfer learning results to improve significantly. We show that Transfer learning is capable to create powerful models for genotype-phenotype predictions in large, well-studied populations and fine-tune these models to populations were data is sparse.

Identification of the cuproptosis-related molecular subtypes and an immunotherapy prognostic model in hepatocellular carcinoma

BACKGROUND

Cuproptosis, a newly discovered mode of cell death, has been less studied in hepatocellular carcinoma (HCC). Exploring the molecular characteristics of different subtypes of HCC based on cuproptosis-related genes (CRGs) is meaningful to HCC. In addition, immunotherapy plays a pivotal role in treating HCC. Exploring the sensitivity of immunotherapy and building predictive models are critical for HCC.

METHODS

The 357 HCC samples from the TCGA database were classified into three subtypes, Cluster 1, Cluster 2, and Cluster 3, based on the expression levels of ten CRGs genes using consensus clustering. Six machine learning algorithms were used to build models that identified the three subtypes. The molecular features of the three subtypes were analyzed and compared from some perspectives. Moreover, based on the differentially expressed genes (DEGs) between Cluster 1 and Cluster 3, a prognostic scoring model was constructed using LASSO regression and Cox regression, and the scoring model was used to predict the efficacy of immunotherapy in the IMvigor210 cohort.

RESULTS

Cluster 3 had the worst overall survival compared to Cluster 1 and Cluster 2 (P = 0.0048). The AUC of the Catboost model used to identify Cluster 3 was 0.959. Cluster 3 was significantly different from the other two subtypes in gene mutation, tumor mutation burden, tumor microenvironment, the expression of immune checkpoint inhibitor genes and N⁶-methyladenosine regulatory genes, and the sensitivity to sorafenib. We believe Cluster 3 is more sensitive to immunotherapy from the above analysis results. Therefore, based on the DEGs between Cluster 1 and Cluster 3, we obtained a 7-gene scoring prognostic model, which achieved meaningful results in predicting immunotherapy efficacy in the IMvigor210 cohort (P = 0.013).

CONCLUSIONS

Our study provides new ideas for molecular characterization and immunotherapy of HCC from machine learning and bioinformatics. Moreover, we successfully constructed a prognostic model of immunotherapy.

Accurate classification of white blood cells by coupling pre-trained ResNet and DenseNet with SCAM mechanism

BACKGROUND

Via counting the different kinds of white blood cells (WBCs), a good quantitative description of a person's health status is obtained, thus forming the critical aspects for the early treatment of several diseases. Thereby, correct classification of WBCs is crucial. Unfortunately, the manual microscopic evaluation is complicated, time-consuming, and subjective, so its statistical reliability becomes limited. Hence, the automatic and accurate identification of WBCs is of great benefit. However, the similarity between WBC samples and the imbalance and insufficiency of samples in the field of medical computer vision bring challenges to intelligent and accurate classification of WBCs. To tackle these challenges, this study proposes a deep learning framework by coupling the pre-trained ResNet and DenseNet with SCAM (spatial and channel attention module) for accurately classifying WBCs.

RESULTS

In the proposed network, ResNet and DenseNet enables information reusage and new information exploration, respectively, which are both important and compatible for learning good representations. Meanwhile, the SCAM module sequentially infers attention maps from two separate dimensions of space and channel to emphasize important information or suppress unnecessary information, further enhancing the representation power of our model for WBCs to overcome the limitation of sample similarity. Moreover, the data augmentation and transfer learning techniques are used to handle the data of imbalance and insufficiency. In addition, the mixup approach is adopted for modeling the vicinity relation across training samples of different categories to increase the generalizability of the model. By comparing with five representative networks on our developed LDWBC dataset and the publicly available LISC, BCCD, and Raabin WBC datasets, our model achieves the best overall performance. We also implement the occlusion testing by the gradient-weighted class activation mapping (Grad-CAM) algorithm to improve the interpretability of our model.

CONCLUSION

The proposed method has great potential for application in intelligent and accurate classification of WBCs.

Enhancement of Image Classification Using Transfer Learning and GAN-Based Synthetic Data Augmentation

Plastic bottle recycling has a crucial role in environmental degradation and protection. Position and background should be the same to classify plastic bottles on a conveyor belt. The manual detection of plastic bottles is time consuming and leads to human error. Hence, the automatic classification of plastic bottles using deep learning techniques can assist with the more accurate results and reduce cost. To achieve a considerably good result using the DL model, we need a large volume of data to train. We propose a GAN-based model to generate synthetic images similar to the original. To improve the image synthesis quality with less training time and decrease the chances of mode collapse, we propose a modified lightweight-GAN model, which consists of a generator and a discriminator with an auto-encoding feature to capture essential parts of the input image and to encourage the generator to produce a wide range of real data. Then a newly designed weighted average ensemble model based on two pre-trained models, inceptionV3 and xception, to classify transparent plastic bottles obtains an improved classification accuracy of 99.06%.

TLSCA-SVM Fault Diagnosis Optimization Method Based on Transfer Learning

In fault-diagnosis classification, a pressing issue is the lack of target-fault samples. Obtaining fault data requires a great amount of time, energy and financial resources. These factors affect the accuracy of diagnosis. To address this problem, a novel fault-diagnosis-classification optimization method, namely TLSCA-SVM, which combines the sine cosine algorithm and support vector machine (SCA-SVM) with transfer learning, is proposed here. Considering the availability of fault data, this thesis uses the data generated by analog circuits from different faults for analysis. Firstly, the data signal is collected from different faults of the analog circuit, and then the characteristic data are extracted from the data signals by the wavelet packets. Secondly, to employ the principal component analysis (PCA) reduces the feature-value dimension. Lastly, as an auxiliary condition, the error-penalty item is added to the objective function of the SCA-SVM classifier to construct an innovative fault-diagnosis model namely TLSCA-SVM. Among them, the Sallen–Key bandpass filter circuit and the CSTV filter circuit are used to provide the data for horizontal- and vertical-contrast classification results. Comparing the SCA with the five optimization algorithms, it is concluded that the performance of SCA optimization parameters has certain advantages in the classification accuracy and speed. Additionally, to prove the superiority of the SCA-SVM classification algorithm, the five classification algorithms are compared with the SCA-SVM algorithm. Simulation results showed that the SCA-SVM classification has higher precision and a faster response time compared to the others. After adding the error penalty term to SCA-SVM, TLSCA-SVM requires fewer fault samples to process fault diagnosis. Ultimately, the method which is proposed could not only perform fault diagnosis effectively and quickly, but also could run effectively to achieve the effect of transfer learning in the case of less failure data.

An Effective Multi-Scale Feature Network for Detecting Connector Solder Joint Defects

With the rapid development of industry, people’s requirements for the functionality, stability, and safety of electronic products are becoming higher and higher. As an important medium for power supply and information transmission functions of electronic products, high-quality soldering of cables and connectors ensures that the devices can operate normally. In this paper, we propose a multi-level feature detection network based on multi-level feature maps fusion and feature enhancement for detecting connector solder joints, classifying and locating qualified solder joints, and detecting seven common defective solder joints. This paper proposes a new feature map up-sampling algorithm and introduces a feature enhancement module, which better preserves the semantic information of higher-level feature maps, while at the same time enhancing the fused feature maps and weakening the effect of noise. Through comparison experiments, the mAP of the network proposed in this paper reaches 0.929 and the top-1 accuracy reaches 92%. The detection capability of each type of solder joint is greatly improved compared with the effect of other networks, which can assist engineers in the detection of weld joint quality and thus reduce the workload.