Mapping Data to Deep Understanding: Making the Most of the Deluge of SARS-CoV-2 Genome Sequences

Sokhansanj, Bahrad A.; Rosen, Gail

doi:10.1128/msystems.00035-22

Cited by 6 publications

(9 citation statements)

References 98 publications

(95 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The relative success of a mixed effect modeling approach suggests that refining the modeling of group level random effects or otherwise incorporate hidden variables are necessary to account for structural issues in the data. Moreover, having established a proof of concept in this study using logistic regression and boosted decision trees, future work can explore the potential application of deep learning methods, which have proven to be highly useful to genetic sequence to function modeling in other contexts [49] , [145] , [146] , [147] .…”

Section: Discussionmentioning

confidence: 99%

“…The challenges of country-level variation are heightened by substantial regional imbalances in the GISAID patient data set. The entire GISAID database is fundamentally biased towards Europe, North America, and select countries in Asia and elsewhere, with over half the sample originating in either the United Kingdom or United States as of January 2022 [49] . Within the subset of data with patient status metadata, the biases are similarly idiosyncratic; for example, over 40% of the training and test samples shown in this paper were obtained from France.…”

Section: Discussionmentioning

confidence: 99%

“…(2) While unprecedented numbers of SARS-CoV-2 sequences have been deposited, that still represents only a sampling of viral infections worldwide. For example, as of April 2022, over half of all sequences in GISAID are from the United States and United Kingdom [49] . The set of sequences with GISAID patient metadata which we were able to curate (excluding illegible or unknown metadata fields) are an even smaller subsample.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Predicting COVID-19 disease severity from SARS-CoV-2 spike protein sequence by mixed effects machine learning

Sokhansanj

Rosen

2022

Computers in Biology and Medicine

Self Cite

View full text Add to dashboard Cite

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Predicting COVID-19 disease severity from SARS-CoV-2 spike protein sequence by mixed effects machine learning

Sokhansanj

Rosen

2022

Computers in Biology and Medicine

Self Cite

View full text Add to dashboard Cite

“…In particular, as COVID-19 shifts from a pandemic to an endemic state, emerging genetic variants of the virus may have different health burdens and require different public health responses [ 2 ]. As a result, there is a critical need for modeling methods capable of predicting the risk of severe disease burden on hospitals and vulnerable populations as a result of continued viral evolution [ 3 ].…”

Section: Introductionmentioning

confidence: 99%

Interpretable and Predictive Deep Neural Network Modeling of the SARS-CoV-2 Spike Protein Sequence to Predict COVID-19 Disease Severity

Sokhansanj

Zhao

Rosen

2022

Biology

Self Cite

View full text Add to dashboard Cite

Through the COVID-19 pandemic, SARS-CoV-2 has gained and lost multiple mutations in novel or unexpected combinations. Predicting how complex mutations affect COVID-19 disease severity is critical in planning public health responses as the virus continues to evolve. This paper presents a novel computational framework to complement conventional lineage classification and applies it to predict the severe disease potential of viral genetic variation. The transformer-based neural network model architecture has additional layers that provide sample embeddings and sequence-wide attention for interpretation and visualization. First, training a model to predict SARS-CoV-2 taxonomy validates the architecture’s interpretability. Second, an interpretable predictive model of disease severity is trained on spike protein sequence and patient metadata from GISAID. Confounding effects of changing patient demographics, increasing vaccination rates, and improving treatment over time are addressed by including demographics and case date as independent input to the neural network model. The resulting model can be interpreted to identify potentially significant virus mutations and proves to be a robust predctive tool. Although trained on sequence data obtained entirely before the availability of empirical data for Omicron, the model can predict the Omicron’s reduced risk of severe disease, in accord with epidemiological and experimental data.

show abstract

“…Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the causative virus of coronavirus disease 2019 (COVID- 19), emerged at the end of 2019, burdening both the global economy and public health [1][2][3][4]. Next-generation sequencing has provided an unprecedented opportunity to monitor the COVID-19 pandemic in real-time [5,6]. During the pandemic, vast amounts of SARS-CoV-2 genome sequences have been accumulated at ever-growing rates and shared in the public database.…”

Section: Introductionmentioning

confidence: 99%

Towards Efficient and Accurate SARS-CoV-2 Genome Sequence Typing Based on Supervised Learning Approaches

Miao

Clercq

2022

Microorganisms

View full text Add to dashboard Cite

Despite the active development of SARS-CoV-2 surveillance methods (e.g., Nextstrain, GISAID, Pangolin), the global emergence of various SARS-CoV-2 viral lineages that potentially cause antiviral and vaccine failure has driven the need for accurate and efficient SARS-CoV-2 genome sequence classifiers. This study presents an optimized method that accurately identifies the viral lineages of SARS-CoV-2 genome sequences using existing schemes. For Nextstrain and GISAID clades, a template matching-based method is proposed to quantify the differences between viral clades and to play an important role in classification evaluation. Furthermore, to improve the typing accuracy of SARS-CoV-2 genome sequences, an ensemble model that integrates a combination of machine learning-based methods (such as Random Forest and Catboost) with optimized weights is proposed for Nextstrain, Pangolin, and GISAID clades. Cross-validation is applied to optimize the parameters of the machine learning-based method and the weight settings of the ensemble model. To improve the efficiency of the model, in addition to the one-hot encoding method, we have proposed a nucleotide site mutation-based data structure that requires less computational resources and performs better in SARS-CoV-2 genome sequence typing. Based on an accumulated database of >1 million SARS-CoV-2 genome sequences, performance evaluations show that the proposed system has a typing accuracy of 99.879%, 97.732%, and 96.291% for Nextstrain, Pangolin, and GISAID clades, respectively. A single prediction only takes an average of <20 ms on a portable laptop. Overall, this study provides an efficient and accurate SARS-CoV-2 genome sequence typing system that benefits current and future surveillance of SARS-CoV-2 variants.

show abstract

Mapping Data to Deep Understanding: Making the Most of the Deluge of SARS-CoV-2 Genome Sequences

Abstract: Next-generation sequencing has been essential to the global response to the COVID-19 pandemic. As of January 2022, nearly 7 million severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) sequences are available to researchers in public databases.

Cited by 6 publications

References 98 publications

Predicting COVID-19 disease severity from SARS-CoV-2 spike protein sequence by mixed effects machine learning

Predicting COVID-19 disease severity from SARS-CoV-2 spike protein sequence by mixed effects machine learning

Interpretable and Predictive Deep Neural Network Modeling of the SARS-CoV-2 Spike Protein Sequence to Predict COVID-19 Disease Severity

Towards Efficient and Accurate SARS-CoV-2 Genome Sequence Typing Based on Supervised Learning Approaches

Contact Info

Product

Resources

About