v4.5.1

MFEprimer v4.5.1

Query -k auto-detect from the index, plus index robustness fixes from late July.

Highlights

• Query -k auto-detect (mfeprimer / spec / search)
  Omit -k to read k from each -d index (.primerqc.bin header; legacy: first .primerqc.fai length). An explicit -k must match every -d index, or the run errors — no more silent 0 binding sites when query k > index k. Multiple -d parts must share the same k. Motivated by large / repeat-rich assemblies (issue #34).

• Index robustness

  • Atomic publish of .primerqc.bin (.partial → fsync → rename) so a killed build cannot leave a truncated index that looks valid.
  • Phase-2 disk spill under memory pressure (sequential bucket encode).
  • Stricter decode / OpenIndex / migrate / verify / indexcmp (reject truncated postings; k capped at 15; migrate prefers source k).

• Fix: forward binding start display
  After the exclusive-bound coordinate change for SNP matching, the report template now shows the correct 1-based inclusive forward binding start.

Docs

• Guide: Large / repeat-rich genomes — choosing index k, split FASTA + multi--d, --bind-amp-only, reading PBN/MBN vs amplicon count.

Quick start

mfeprimer index -i genome.fa -k 13 -c 32 -f
mfeprimer spec -i primers.fa -d genome.fa -o out.spec --bind-amp-only -b -j
# omit -k: auto from index