Toward Efficient Genomic Compression Using Channel Codes For Joint Alignment And Reconstruction
summary
Proposes a low-complexity compression pipeline that keeps the reference only at the cloud (decoder-only reference) and encodes short samples so alignment and reconstruction share bits; formalizes a Single Deletion Multiple Substitutions (SDMS model), invents a shift-compensating distance computable in linear time for one deletion, derives a closed-form false-alignment distribution showing false hits drop from >50% to ~10% at a practical threshold, and reconstructs by expanding a small deletion-candidate list and running substitution-only decoders (encoder stays a simple matrix–vector product).
AI-written summary · about
chapters (10)
- 0:00Genomic data compression challenges
- 3:58Genomic compression necessity
- 7:43Coding theory foundations
- 12:34Alignment in reconstruction
- 15:10Substitution and deletion errors
- 18:32Alignment distance analysis
- 22:09Reusing alignment for reconstruction
- 26:51Proposed coding scheme summary
- 27:15Future directions
- 30:52Q&A
related talks
10mRevisiting Network Coding for Warm Blob Storage
11mExploiting Combined Locality for Wide-Stripe Erasure Coding in Distributed Storage
31mMulti-version Coding and Shared Memory Emulation
15mLESS is More for I/O-Efficient Repairs in Erasure-Coded Storage
15mC2DN: How to Harness Erasure Codes at the Edge for Efficient Content Delivery
01h 14mErasure Coding at Scale
15mSelectiveEC: Selective Reconstruction in Erasure-coded Storage Systems
17mFrom Channel Assignment to Subgraph Isomorphism