๐Ÿ  Portal Home
VBASE3 Logo

๐Ÿ“Š Expressed Antibody Repertoire Diversity

๐Ÿ“ฆ Zenodo Data โ†— ๐Ÿ  Portal Home
Multi-Dataset Manifold: Complete V(D)J Sequences, Hypervariable CDR3 Loops & Paired Single-Cell BCRs
๐Ÿ  Portal Home ๐Ÿ“Š Repertoire ๐Ÿ›ก๏ธ BCRbase ๐Ÿงฌ TCR Rep
โšก Client WASM: Active (In-Browser)
-Visible Clones
-Datasets Loaded
-V-Families
Single-CellVH+VL & VHH
๐Ÿ“‚ Repertoire Dataset Manager
Toggle active reference repertoires or upload custom single-cell / bulk AIRR datasets to project onto the manifold.
Active Datasets
Curated Baseline Presets
โšก WASM Model:
๐Ÿ”’ 100% Local Browser CPU
๐Ÿ“ฅ
Drag & drop FASTQ / FASTA / AIRR TSV here
Supports assembled NGS FASTQ reads, FASTA contigs & AIRR tables
๐Ÿ”ฌ Repertoire Clone Inspector
Click on any repertoire point on the manifold to inspect its full antibody sequence, hypervariable CDR3 loop, and paired light chain partner.
Dataset: - Organism: - Tech: - Chain: - Family: -
Variable Gene (V)
-
Diversity Element (D)
-
Joining Element (J)
-
CDR3 Loop
-
-
AMINO ACID SEQUENCE:
Click a point to display sequence...
โšก VBASE3 Performance Benchmark: Client WASM vs. Server Native Rust
Side-by-side benchmark executing identical antibody query sets across the Browser WebAssembly Engine (single-threaded client sandbox) and Server Native Rust (Apple M4 Pro host with multi-core Rayon parallelization and NEON 128-bit SIMD).
โ„น๏ธ
Release Note: Benchmark disabled in the WASM only version. When running the web pages solely in a web browser without the local native Rust server daemon (dnaplot_server), live server benchmarking cannot connect to /api/benchmark. Canonical reference metrics measured on an Apple M4 Pro (16-Core Rayon SIMD, 509x speedup) are displayed below and documented in Manuscript Appendix Figure App 11.
โšก Client WebAssembly (In-Browser) V8 / WebKit
Architecture wasm32-unknown-unknown
Concurrency 1 Thread (Browser Sandbox)
Alignment Throughput 17.8 seqs / sec
Alignment Latency 56.21 ms / seq
PaCMAP Embedding (100 seqs) 7.5 ms (Single-Core)
Network Overhead 0 ms (100% Local)
๐Ÿš€ Server Native Rust Apple M4 Pro
Architecture & SIMD aarch64 โ€ข NEON 128-bit
Concurrency 16 Cores (Rayon Work-Stealing)
Alignment Throughput 9,064 queries / sec
Alignment Latency 110.33 ยตs / query
PaCMAP Embedding (100 seqs) 153.75 ms (650.4 seqs/s)
Zero-Shift Pairwise Rate 19.1 M comparisons/sec
Comparative Speedup: Server Native Rust on Apple M4 Pro achieves 509x higher throughput (19.1M pairwise comparisons/sec).
๐Ÿš€ 509x FASTER (M4 Pro)