VLPIM README

Back to Home GitHub
Overview

VLPIM is a comprehensive computational workflow for modulating the immunogenicity of virus-like particles through epitope identification, MHC-II binding evaluation, and structure analysis.

  • Epitope prediction: Candidate epitopes predicted by NetMHCIIpan
  • Immunogenicity analysis: Analyzing immunogenicity shift of candidate epitopes predicted by NetMHCIIpan
  • Structural superposition: Structures comparative analysis between AF3 predicted structure and wild-type structure
Workflow
Workflow of VLPIM
Figure 1. Workflow of VLPIM.
  • Step 1: Dominant and subdominant antigenic epitopes are first identified using immune epitope databases and the NetMHCIIpan prediction model.
  • Step 2: These epitopes provide the basis for generating a mutant protein library through ProteinMPNN-based deep learning methods.
  • Step 3: Structural and immunogenicity predictions are then applied to evaluate the effects of mutations.
  • Step 4: Candidate variants are further filtered using AF3 outputs, Cα RMSD, and Rosetta.
  • Wet lab evaluation: Finally, the most promising mutants could be validated in experimental studies to assess their physicochemical properties and immunogenicity. AF3, Alphafold3; RMSD, root mean square deviation.
Usage Limits

The following practical constraints and parameter limits apply to the VLPIM web server:

Sequence and Input Limits
  • VLP sequence length: Recommended up to 10,000 amino acids. Very long sequences may experience slower processing times.
  • NetMHCIIpan file: Standard output format expected. Large files (>10MB) may require longer processing time.
  • PDB structure files: Recommended for structures with up to 5,000 residues per chain for optimal performance.
Parameter Ranges
  • Epitope number: 1-50 epitopes (default: 10)
  • Epitope target length: 9-15 amino acids (default: 15 aa)
  • RMSD cutoff: 0.5-500 Å (suggested range: 2-300 Å)
  • RMSD max cycles: 1-20 iterations (default: 5 cycles)
Runtime Expectations
  • Epitope identification: Typically completes within 5-30 seconds depending on sequence length and number of epitopes requested.
  • RMSD calculation: Usually completes within 10-60 seconds for typical protein structures. Iterative alignment may take longer with higher cycle counts.
  • API requests: Timeout set to 10 seconds for external API calls.
General Guidelines
  • For best performance, use sequences and structures of moderate size (< 5,000 residues).
  • Large batch analyses may require multiple sessions or splitting input data.
  • Browser-based computation is performed client-side; ensure sufficient browser memory for large datasets.
  • Results are computed in real-time and not stored on the server.
Privacy and Analytics
  • Sequence and structure analysis is executed locally in the browser.
  • Anonymous traffic analytics (GA4/Looker Studio dashboard) is optional and disabled by default.
  • When analytics is enabled by the user, page-view statistics may be sent to Google Analytics.
Practical Website Example (Built-in Demo Data)

Follow the actual UI flow: load examples -> run calculations -> review outputs and export files.

This walkthrough is aligned with the current index.html UI and uses built-in sample files for a complete end-to-end run.

DP_P03146_NetMHCIIpan.xls seq-a .csv 6htx.pdb
Step 1: Epitope identify

Open index.html, in Step 1 click Load example, then run Epitopes identification.

Expected output: identified epitope table and Export extended epitopes file selected_epitopes.csv.
Step 2: Immunogenicity analyze (Step 4.1)

Go to Step 4.1, load seq-a .csv, select mode (enhance/reduce), then compute Sim score.

Expected output: peptide score table and Export CSV combined file (e.g., sim_im_reduce_combined.csv).
Step 3: Structural superposition (Step 4.2)

Go to Step 4.2, use Load example 1 or Load example 2, choose Kabsch/Iterative/TM-align, and click Calculate RMSD.

Expected output: structure alignment result card with method metrics (RMSD and TM-score when TM-align is selected).
Quick check: All three modules should complete without errors and produce visible outputs (tables/cards) plus downloadable CSV files where available.
Figure 2: Practical workflow aligned with current UI Website sequence: load example -> run calculation -> validate/export Step 1 Epitope identify Input: DP_P03146... Output: selected_epitopes.csv Step 4.1 Sim analysis Input: seq-a .csv Output: sim_im_*.csv Step 4.2 RMSD superposition Input: 6htx.pdb demo Output: RMSD result card Action checklist 1. Use Load example / Load example 1 / Load example 2 buttons 2. Run Epitopes identification, Compute S im scores, and Calculate RMSD 3. Verify result tables/cards and use Export buttons 4. Re-run with mode/method switches for consistency check
Figure 2. Function usage map based on built-in examples, updated to match the current page structure and button labels.
Legend: Sim indicates immunogenicity score analysis; RMSD indicates root mean square deviation; TM-score is shown when TM-align is selected.
© 2025 VLPIM. MIT License.