Use VaxPress Online

VaxPress: A Powerhouse for mRNA Vaccine Optimization

VaxPress is an advanced, automated pipeline designed to optimize synonymous codon selections, significantly enhancing the fitness of coding sequences (CDS) for mRNA vaccines and therapeutics.

Traditionally, designing high-yielding, stable mRNA sequences required manual tinkering or heavy command-line processing to balance conflicting sequence features. VaxPress solves this by implementing a highly customizable genetic algorithm. It refines wild-type sequences across a multitude of critical biochemical metrics, producing mRNA candidates optimized for cell-based expression and manufacturing stability.

The Core Optimization Engine

VaxPress operates through an evolutionary iterative process that continuously filters for optimal sequence traits:

• Genetic Algorithm: The program initializes a collection of randomly mutated child sequences (configured via population size). Scoring functions evaluate each sequence, and a selection process filters out the fittest variants ("survivors") to produce the next generation's offspring under a pre-defined mutation rate. This continues across hundreds to thousands of iterations until the population converges at peak fitness.

• Linear Combination Objective Function: Total sequence fitness is calculated as a weighted linear combination of individual metrics:

  

• Adaptive Mutation Rate (Winddown): To prevent optimization from stalling, VaxPress uses a winddown trigger. If the fitness score plateaus over a set number of cycles, it automatically scales down the mutation rate to finely hone the existing top-performing sequences.

Key Optimization Features & Metrics

VaxPress evaluates and fine-tunes mRNA sequences by adjusting the specific weights of individual scoring functions (which can be actively dialed up or disabled entirely by setting weights to 0).

Advanced Workflows: Integrating LinearDesign

For the highest-quality outputs, VaxPress seamlessly integrates with LinearDesign to establish an industry-leading hybrid optimization pipeline.

[Protein / mRNA Input] 
         │
         ▼
 ┌───────────────┐
 │ LinearDesign  │ ──► Establishes global near-optimal MFE & CAI baseline
 └───────────────┘
         │
         ▼
 ┌───────────────┐
 │   VaxPress    │ ──► Eliminates Tandem Repeats, refines Start Codon loops,
 └───────────────┘     tunes Uridine Count, and maximizes In-Cell Stability
         │
         ▼
[Final Optimized CDS]

[Protein / mRNA Input] 
         │
         ▼
 ┌───────────────┐
 │ LinearDesign  │ ──► Establishes global near-optimal MFE & CAI baseline
 └───────────────┘
         │
         ▼
 ┌───────────────┐
 │   VaxPress    │ ──► Eliminates Tandem Repeats, refines Start Codon loops,
 └───────────────┘     tunes Uridine Count, and maximizes In-Cell Stability
         │
         ▼
[Final Optimized CDS]

[Protein / mRNA Input] 
         │
         ▼
 ┌───────────────┐
 │ LinearDesign  │ ──► Establishes global near-optimal MFE & CAI baseline
 └───────────────┘
         │
         ▼
 ┌───────────────┐
 │   VaxPress    │ ──► Eliminates Tandem Repeats, refines Start Codon loops,
 └───────────────┘     tunes Uridine Count, and maximizes In-Cell Stability
         │
         ▼
[Final Optimized CDS]

Why Initialize with LinearDesign?

LinearDesign handles the heavy structural architecture by instantly determining a sequence with a virtually minimum free energy secondary structure and optimal CAI among all possible codon combinations.

VaxPress then picks up where LinearDesign leaves off, taking that pre-refined baseline and optimizing the additional multidimensional criteria (such as sequence repetitions, uridine count, and local cell dynamics) that LinearDesign's algorithm does not natively consider.

Conservative Start Strategy: Sequences pulled straight from LinearDesign often feature suboptimal structures at the N-terminus. VaxPress counteracts this by utilizing a specialized focus mode that isolates and mutates only the start codon region during initial generations, preventing the rest of the sequence body from losing its pristine MFE baseline structure.

What is Tamarind Bio?

Tamarind Bio is a pioneering, no-code bioinformatics platform built to democratize access to powerful computational tools for life scientists and researchers worldwide. Recognizing that cutting-edge machine learning models and evolutionary algorithms are frequently bottlenecked by intricate software dependencies, high-performance computing management, and rigid command-line interfaces, Tamarind provides an intuitive, web-based ecosystem.

By abstracting away the underlying technical complexities of cloud infrastructure, Tamarind Bio empowers molecular biologists, biochemists, and pharma teams to focus entirely on their scientific questions. The platform delivers secure data management, automated computational scaling, and highly interactive visual data reports—bringing state-of-the-art computational biology right to your web browser.

How to Use VaxPress on Tamarind Bio

Running a comprehensive, multi-metric mRNA vaccine optimization through VaxPress normally requires dedicated Linux clusters. On Tamarind Bio, the workflow is streamlined into a seamless, no-code graphical process:

1. Access the Tool and Upload Sequence:

Log into your Tamarind Bio workspace and open the VaxPress Online tool. Input your target sequence into the interface using a standard FASTA file format. You can provide an existing wild-type mRNA CoDing Sequence (CDS) or upload a raw protein sequence directly (enabling the protein sequence toggle).

2. Configure Optimization Iterations and Parallelization:

Set the depth of your evolutionary search. While the absolute default is low, it is highly recommended to allocate at least 500 to 1,500 iterations for thorough sequence convergence. Choose your parallelization capacity (CPU cores) to optimize secondary structure prediction speeds.

3. Adjust Fitness Function Weight Schemes:

Tailor the objective function to your specific vaccine requirements. Use the interface sliders to increase or decrease feature weights. For example, if you want an ultra-stable construct, dial up the MFE Weight. To erase manufacturing hurdles found in purely CAI-driven designs, set a high Repeats Weight to purge tandem repeats.

4. Enable LinearDesign Initialization (Optional):

Toggle on the LinearDesign initialization feature. Enter your desired Lambda ($\lambda$) parameter value (typically between 0.5 and 4.0) to balance global structural energy against codon bias. Define your conservative start range to shield the rest of the sequence from early high-mutation-rate disruptions.

5. Launch and Download Interactive Reports:

Execute the optimization job. Once completed, your output directory provides full access to your refined data files: download the finalized best-sequence.fasta, inspect the generational tracking inside checkpoints.tsv, or open the fully interactive report.html to visualize the predicted mRNA secondary structures and track metric convergence plots over time.

Source