scholarly journals Evolutionarily informed machine learning enhances the power of predictive gene-to-phenotype relationships

2021 ◽  
Vol 12 (1) ◽  
Author(s):  
Chia-Yi Cheng ◽  
Ying Li ◽  
Kranthi Varala ◽  
Jessica Bubert ◽  
Ji Huang ◽  
...  

AbstractInferring phenotypic outcomes from genomic features is both a promise and challenge for systems biology. Using gene expression data to predict phenotypic outcomes, and functionally validating the genes with predictive powers are two challenges we address in this study. We applied an evolutionarily informed machine learning approach to predict phenotypes based on transcriptome responses shared both within and across species. Specifically, we exploited the phenotypic diversity in nitrogen use efficiency and evolutionarily conserved transcriptome responses to nitrogen treatments across Arabidopsis accessions and maize varieties. We demonstrate that using evolutionarily conserved nitrogen responsive genes is a biologically principled approach to reduce the feature dimensionality in machine learning that ultimately improved the predictive power of our gene-to-trait models. Further, we functionally validated seven candidate transcription factors with predictive power for NUE outcomes in Arabidopsis and one in maize. Moreover, application of our evolutionarily informed pipeline to other species including rice and mice models underscores its potential to uncover genes affecting any physiological or clinical traits of interest across biology, agriculture, or medicine.

F1000Research ◽  
2019 ◽  
Vol 7 ◽  
pp. 702
Author(s):  
Pedro Morell Miranda ◽  
Francesca Bertolini ◽  
Haja N. Kadarmideen

Background: Inflammatory bowel disease (IBD) is a group of chronic diseases related to inflammatory processes in the digestive tract generally associated with an immune response to an altered gut microbiome in genetically predisposed subjects. For years, both researchers and clinicians have been reporting increased rates of anxiety and depression disorders in IBD, and these disorders have also been linked to an altered microbiome. However, the underlying pathophysiological mechanisms of comorbidity are poorly understood at the gut microbiome level. Methods: Metagenomic and metatranscriptomic data were retrieved from the Inflammatory Bowel Disease Multi-Omics Database. Samples from 70 individuals that had answered to a self-reported depression and anxiety questionnaire were selected and classified by their IBD diagnosis and their questionnaire results, creating six different groups. The cross-validation random forest algorithm was used in 90% of the individuals (training set) to retain the most important species involved in discriminating the samples without losing predictive power. The validation set that represented the remaining 10% of the samples equally distributed across the six groups was used to train a random forest using only the species selected in order to evaluate their predictive power. Results: A total of 24 species were identified as the most informative in discriminating the 6 groups. Several of these species were frequently described in dysbiosis cases, such as species from the genus Bacteroides and Faecalibacterium prausnitzii. Despite the different compositions among the groups, no common patterns were found between samples classified as depressed. However, distinct taxonomic profiles within patients of IBD depending on their depression status were detected. Conclusions: The machine learning approach is a promising approach for investigating the role of microbiome in IBD and depression. Abundance and functional changes in these species suggest that depression should be considered as a factor in future research on IBD.


Author(s):  
Ayal B. Gussow ◽  
Sergey A. Shmakov ◽  
Kira S. Makarova ◽  
Yuri I. Wolf ◽  
Joseph Bondy-Denomy ◽  
...  

AbstractBacteria and archaea evolve under constant pressure from numerous, diverse viruses and thus have evolved multiple defense systems. The CRISPR-Cas are adaptive immunity systems that have been harnessed for the development of the new generation of genome editing and engineering tools. In the incessant host-parasite arms race, viruses evolved multiple anti-defense mechanisms including numerous, diverse anti-CRISPR proteins (Acrs) that can inhibit CRISPR-Cas and therefore have enormous potential for application as modulators of genome editing tools. Most Acrs are small, highly variable proteins which makes their prediction a formidable task. We developed a machine learning approach for comprehensive Acr prediction. The model showed high predictive power when tested against an unseen test set that included several families of recently discovered Acrs and was employed to predict 2,500 novel candidate Acr families. An examination of the top candidates confirms that they possess typical Acr features. One of the top candidates was independently tested and found to possess anti-CRISPR activity (AcrIIA12). We provide a web resource (http://acrcatalog.pythonanywhere.com/) to access the predicted Acrs sequences and annotation. The results of this analysis expand the repertoire of predicted Acrs almost by two orders of magnitude and provide a rich resource for experimental Acr discovery.


Cancers ◽  
2021 ◽  
Vol 13 (17) ◽  
pp. 4468
Author(s):  
Seokjin Haam ◽  
Jae-Ho Han ◽  
Hyun Woo Lee ◽  
Young Wha Koh

Using a machine learning approach with a gene expression profile, we discovered a tumor nonimmune-microenvironment-related gene expression signature, including extracellular matrix (ECM) remodeling, epithelial–mesenchymal transition (EMT), and angiogenesis, that could predict brain metastasis (BM) after the surgical resection of 64 lung adenocarcinomas (LUAD). Gene expression profiling identified a tumor nonimmune-microenvironment-related 17-gene expression signature that significantly correlated with BM. Of the 17 genes, 11 were ECM-remodeling-related genes. The 17-gene expression signature showed high BM predictive power in four machine learning classifiers (areas under the receiver operating characteristic curve = 0.845 for naïve Bayes, 0.849 for support vector machine, 0.858 for random forest, and 0.839 for neural network). Subgroup analysis revealed that the BM predictive power of the 17-gene signature was higher in the early-stage LUAD than in the late-stage LUAD. Pathway enrichment analysis showed that the upregulated differentially expressed genes were mainly enriched in the ECM–receptor interaction pathway. The immunohistochemical expression of the top three genes of the 17-gene expression signature yielded similar results to NanoString tests. The tumor nonimmune-microenvironment-related gene expression signatures found in this study are important biological markers that can predict BM and provide patient-specific treatment options.


10.2196/24572 ◽  
2021 ◽  
Vol 9 (2) ◽  
pp. e24572
Author(s):  
Juan Carlos Quiroz ◽  
You-Zhen Feng ◽  
Zhong-Yuan Cheng ◽  
Dana Rezazadegan ◽  
Ping-Kang Chen ◽  
...  

Background COVID-19 has overwhelmed health systems worldwide. It is important to identify severe cases as early as possible, such that resources can be mobilized and treatment can be escalated. Objective This study aims to develop a machine learning approach for automated severity assessment of COVID-19 based on clinical and imaging data. Methods Clinical data—including demographics, signs, symptoms, comorbidities, and blood test results—and chest computed tomography scans of 346 patients from 2 hospitals in the Hubei Province, China, were used to develop machine learning models for automated severity assessment in diagnosed COVID-19 cases. We compared the predictive power of the clinical and imaging data from multiple machine learning models and further explored the use of four oversampling methods to address the imbalanced classification issue. Features with the highest predictive power were identified using the Shapley Additive Explanations framework. Results Imaging features had the strongest impact on the model output, while a combination of clinical and imaging features yielded the best performance overall. The identified predictive features were consistent with those reported previously. Although oversampling yielded mixed results, it achieved the best model performance in our study. Logistic regression models differentiating between mild and severe cases achieved the best performance for clinical features (area under the curve [AUC] 0.848; sensitivity 0.455; specificity 0.906), imaging features (AUC 0.926; sensitivity 0.818; specificity 0.901), and a combination of clinical and imaging features (AUC 0.950; sensitivity 0.764; specificity 0.919). The synthetic minority oversampling method further improved the performance of the model using combined features (AUC 0.960; sensitivity 0.845; specificity 0.929). Conclusions Clinical and imaging features can be used for automated severity assessment of COVID-19 and can potentially help triage patients with COVID-19 and prioritize care delivery to those at a higher risk of severe disease.


2020 ◽  
Author(s):  
Juan Quiroz ◽  
Youzhen Feng ◽  
Zhongyuan Cheng ◽  
Dana Rezazadegan ◽  
Pingkang Chen ◽  
...  

Objectives This study aims to develop a machine learning approach for automated severity assessment of COVID-19 patients based on clinical and imaging data. Materials and Methods Clinical data, including demographics, signs, symptoms, comorbidities and blood test results and chest CT scans of 346 patients from two hospitals in the Hubei province, China, were used to develop machine learning models for automated severity assessment of diagnosed COVID-19 cases. We compared the predictive power of clinical and imaging data by testing multiple machine learning models, and further explored the use of four oversampling methods to address the imbalance distribution issue. Features with the highest predictive power were identified using the SHAP framework. Results Targeting differentiation between mild and severe cases, logistic regression models achieved the best performance on clinical features (AUC:0.848, sensitivity:0.455, specificity:0.906), imaging features (AUC:0.926, sensitivity:0.818, specificity:0.901) and the combined features (AUC:0.950, sensitivity:0.764, specificity:0.919). The SMOTE oversampling method further improved the performance of the combined features to AUC of 0.960 (sensitivity:0.845, specificity:0.929). Discussion Imaging features had the strongest impact on the model output, while a combination of clinical and imaging features yielded the best performance overall. The identified predictive features were consistent with findings from previous studies. Oversampling yielded mixed results, although it achieved the best performance in our study. Conclusions This study indicates that clinical and imaging features can be used for automated severity assessment of COVID-19 patients and have the potential to assist with triaging COVID-19 patients and prioritizing care for patients at higher risk of severe cases. [Manuscript last updated on 31 July, 2020]


2020 ◽  
Author(s):  
Juan Carlos Quiroz ◽  
You-Zhen Feng ◽  
Zhong-Yuan Cheng ◽  
Dana Rezazadegan ◽  
Ping-Kang Chen ◽  
...  

BACKGROUND COVID-19 has overwhelmed health systems worldwide. It is important to identify severe cases as early as possible, such that resources can be mobilized and treatment can be escalated. OBJECTIVE This study aims to develop a machine learning approach for automated severity assessment of COVID-19 based on clinical and imaging data. METHODS Clinical data—including demographics, signs, symptoms, comorbidities, and blood test results—and chest computed tomography scans of 346 patients from 2 hospitals in the Hubei Province, China, were used to develop machine learning models for automated severity assessment in diagnosed COVID-19 cases. We compared the predictive power of the clinical and imaging data from multiple machine learning models and further explored the use of four oversampling methods to address the imbalanced classification issue. Features with the highest predictive power were identified using the Shapley Additive Explanations framework. RESULTS Imaging features had the strongest impact on the model output, while a combination of clinical and imaging features yielded the best performance overall. The identified predictive features were consistent with those reported previously. Although oversampling yielded mixed results, it achieved the best model performance in our study. Logistic regression models differentiating between mild and severe cases achieved the best performance for clinical features (area under the curve [AUC] 0.848; sensitivity 0.455; specificity 0.906), imaging features (AUC 0.926; sensitivity 0.818; specificity 0.901), and a combination of clinical and imaging features (AUC 0.950; sensitivity 0.764; specificity 0.919). The synthetic minority oversampling method further improved the performance of the model using combined features (AUC 0.960; sensitivity 0.845; specificity 0.929). CONCLUSIONS Clinical and imaging features can be used for automated severity assessment of COVID-19 and can potentially help triage patients with COVID-19 and prioritize care delivery to those at a higher risk of severe disease.


2021 ◽  
Author(s):  
Michael T.W. McKibben ◽  
Michael S. Barker

Nearly all lineages of land plants have experienced at least one whole genome duplication (WGD) in their history. The legacy of these ancient WGDs is still observable in the diploidized genomes of extant plants. Genes originating from WGD-paleologs-can be maintained in diploidized genomes for millions of years. These paleologs have the potential to shape plant evolution through sub- and neofunctionalization, increased genetic diversity, and reciprocal gene loss among lineages. Current methods for classifying paleologs often rely on only a subset of potential genomic features, have varying levels of accuracy, and often require significant data and/or computational time. Here we developed a supervised machine learning approach to classify paleologs from a target WGD in diploidized genomes across a broad range of different duplication histories. We collected empirical data on syntenic block sizes and other genomic features from 27 plant species each with a different history of paleopolyploidy. Features from these genomes were used to develop simulations of syntenic blocks and paleologs to train a gradient boosted decision tree. Using this approach, Frackify (Fractionation Classify), we were able to accurately identify and classify paleologs across a broad range of parameter space, including cases with multiple overlapping WGDs. We then compared Frackify with other paleolog inference approaches in six species with paleotetraploid and paleohexaploid ancestries. Frackify provides a way to combine multiple genomic features to quickly classify paleologs while providing a high degree of consistency with existing approaches.


Cancers ◽  
2019 ◽  
Vol 11 (10) ◽  
pp. 1562 ◽  
Author(s):  
Maurizio Polano ◽  
Marco Chierici ◽  
Michele Dal Bo ◽  
Davide Gentilini ◽  
Federica Di Cintio ◽  
...  

Immunotherapy by using immune checkpoint inhibitors (ICI) has dramatically improved the treatment options in various cancers, increasing survival rates for treated patients. Nevertheless, there are heterogeneous response rates to ICI among different cancer types, and even in the context of patients affected by a specific cancer. Thus, it becomes crucial to identify factors that predict the response to immunotherapeutic approaches. A comprehensive investigation of the mutational and immunological aspects of the tumor can be useful to obtain a robust prediction. By performing a pan-cancer analysis on gene expression data from the Cancer Genome Atlas (TCGA, 8055 cases and 29 cancer types), we set up and validated a machine learning approach to predict the potential for positive response to ICI. Support vector machines (SVM) and extreme gradient boosting (XGboost) models were developed with a 10×5-fold cross-validation schema on 80% of TCGA cases to predict ICI responsiveness defined by a score combining tumor mutational burden and TGF- β signaling. On the remaining 20% validation subset, our SVM model scored 0.88 accuracy and 0.27 Matthews Correlation Coefficient. The proposed machine learning approach could be useful to predict the putative response to ICI treatment by expression data of primary tumors.


Sign in / Sign up

Export Citation Format

Share Document