Modeling the trajectory of SARS-CoV-2 spike protein evolution in continuous latent space using a neural network and Gaussian process
Bibliographic record
Abstract
Abstract Viral vaccines can lose their efficacy as the genomes of targeted viruses rapidly evolve, resulting in new variants that may evade vaccine-induced immunity. This process is apparent in the emergence of new SARS-CoV-2 variants which have the potential to undermine vaccination efforts and cause further outbreaks. Predictive vaccinology points to a future of pandemic preparedness in which vaccines can be developed preemptively based in part on predictive models of viral evolution. Thus, modeling the trajectory of SARS-CoV-2 spike protein evolution could have value for mRNA vaccine development. Traditionally, in silico sequence evolution has been modeled discretely, while there has been limited investigation into continuous models. Here we present the Viral Predictor for mRNA Evolution (VPRE), an open-source software tool which learns from mutational patterns in viral proteins and models their most statistically likely evolutionary trajectories. We trained a variational autoencoder with real-time and simulated SARS-CoV-2 genome data from Australia to encode discrete spike protein sequences into continuous numerical variables. To simulate evolution along a phylogenetic path, we trained a Gaussian process model with the numerical variables to project spike protein evolution up to five months in advance. Our predictions mapped primarily to a sequence that differed by a single amino acid from the most reported spike protein in Australia within the prediction timeframe, indicating the utility of deep learning and continuous latent spaces for modeling viral protein evolution. VPRE can be readily adapted to investigate and predict the evolution of viruses other than SARS-CoV-2 in temporal, geographic, and lineage-specific pathways.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".