DNA data storage encodes digital information in nucleotide sequences. The channel is unusual: synthesis introduces substitution, insertion, and deletion errors; storage may lose entire oligos (short DNA strands); and the read process (sequencing) adds its own noise. Standard error-correcting codes handle substitutions but not the full error profile.
DNA-MGC+ is a modular codec that separates the encoding pipeline into independently swappable stages: digital-to-quaternary mapping, inner error correction, outer error correction (for oligo loss), and biological constraint enforcement (GC content, homopolymer avoidance). Each stage can be optimized or replaced independently — if a new inner code improves substitution correction, it slots in without redesigning the outer code.
Note (2026-11-14; read against the paper's full text): the essay describes DNA-MGC+ as a modular codec of independently swappable stages that includes a biological-constraint enforcement stage; the word modular appears in the paper only in a citation, and I found no statement about swapping stages. The paper's pipeline is: fragmentation of the binary file, an outer Reed-Solomon code for dropouts, indexing and an inner Marker Guess and Check Plus (MGC+) code whose binary-to-quaternary mapping is unconstrained, then optional filtering for constraints (homopolymer length, GC content, motifs, thermodynamic properties), which the outer code makes possible by generating an excess pool of candidate sequences from which a subset is selected without changing the code rate. It says that Gibbs free energy is more strongly associated with coverage variability than the usually constrained homopolymer length and GC content. The results the essay omits are the paper's actual evidence: in silico decoding up to 24 percent error rates for a 15 KB file, a 200 MB file at 5 percent error, and an in vitro 24 KB file with electrochemical synthesis, compared with selected codecs on sequencing depth, read cost, decoding time and density; the inner/outer redundancy tradeoff is described as a design tradeoff for optimizing the parametrization.
The modularity allows tradeoff navigation. More redundancy in the inner code reduces per-base error rate but reduces net information density; more redundancy in the outer code protects against oligo loss but costs additional oligos. The codec exposes these tradeoffs as tunable parameters rather than fixed design choices. Different applications — archival storage (low read frequency, high reliability needed) versus computational storage (frequent reads, density matters) — select different parameter settings within the same framework.
The structural point: DNA storage is not a single channel but a composition of channels (synthesis, storage, sequencing), and the optimal codec is not a single code but a composition of codes. The modular design matches the channel's structure — each code handles the errors from its corresponding channel segment. The codec decomposes because the channel decomposes, and the factorization isn't a design choice but a recognition of the channel's actual structure.