Back

Topology-aware thermodynamics for DNA probe design under fixed stringency: Retained paired boxes link mismatch placement, nearest-neighbor stability and room-temperature diagnostic specificity

Brukner, I.; Krajinovic, M.

2026-06-11 genomics
10.64898/2026.06.10.731169 bioRxiv
Show abstract

DNA probe specificity is usually screened by mismatch number, melting temperature and full-duplex nearest-neighbor free energy. These scalar quantities are necessary but obscure the spatial organization of pairing: an off-target can retain one long paired box or split the same matches into short fragments. We define thermodynamics without topology as scalar full-duplex nearest-neighbor DeltaG and thermodynamics with topology as redistribution of nearest-neighbor contributions over retained paired boxes. The practical score family begins with edge-corrected effective complementary islands, extends to the count Nbox(L,k) of contiguous boxes of length at least k, and culminates in Zbox,NN(k,T), a retained-box partition weighted by nearest-neighbor stability. Reanalysis of published mismatch-probe datasets shows why the distinction matters: distributed or staggered mismatches are far more disruptive than clustered mismatches at comparable burden; 50-mer mismatch probes show distribution- and temperature-dependent residual signal; maximum perfect-match length explains a substantial fraction of 60-mer mismatch signal ratios; and zipper/partition-function modeling explains position-dependent mismatch discrimination. An Affymetrix fixed-mismatch subset and an ambient-temperature HPV probe panel provide independent control and diagnostic context. The framework offers an auditable rule for fixed-stringency probe design: preserve intended retained continuity, fragment the strongest off-target retained box, then validate in the final buffer and readout.

Matching journals

The top 9 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.