[Alexander-Mitrofanov]

Genomic target extraction previously made the requested region's ends artificial folding boundaries. Add genomic context using IntaRNAsTar's target accessibility base-pair span, then restrict interaction prediction to the original region with --tRegion. Report sequences and all target/seed/structure coordinates remain relative to the requested region.

• Honor tAccL/accL, native zero/window semantics, circular wrapping and physical linear ends across both strands and all extraction modes. Preserve existing spliced CDS and short-truncated-UTR behavior.
• Share context-aware prediction between the main workflow and web exports. Use native indexing for report coordinates, trim CSV target sequences, and normalize E_norm before filtering and statistical processing.
• Include extended sequences and intervals in cache signatures, retain context sidecars during cleanup, and protect their output namespace. Document the behavior and intentional changes to accessibility energies/rankings in the README and changelog.
• Address review feedback by emitting empty fields instead of NA placeholders in published result/web tables while preserving internal missing-value semantics, and authenticate DAVID with rna@informatik.uni-freiburg.de.

Validation: 254 passed, 1 skipped, with COPRARNA_PIPELINE_TESTS=1 and COPRARNA_REFERENCE_TESTS=1, using the release's pinned Python libraries. This includes real IntaRNAsTar 3.4.1 predictions, file/inline settings equivalence, web exports, and extraction from all 17 reference genomes. The skip is the optional browser integration. Independently captured scientific fixtures are unchanged. Source/wheel builds succeeded; 79 accessibility tests passed against the installed wheel outside the source tree, including native predictions. Ruff and git diff --check passed.

Targets codex/issue-4-release-3 as requested in the issue.

Closes #6.

[martin-raden]

@Alexander-Mitrofanov

thanks, please extend:

NA use in output

given the discussion in PatrickRWright#22 it is favorable to use empty strings instead of placeholding NA values within the final output tables to simplify reading by the user and importing into subsequent services

DAVID email address

replace the email address used to access the DAVID service with "rna@informatik.uni-freiburg.de"

[Alexander-Mitrofanov]

Addressed the review feedback in 3adace6: published CopraRNA result and web tables now emit empty fields for unavailable values while internal scientific tables retain missing-value semantics, and DAVID authenticates with rna@informatik.uni-freiburg.de. Release-style validation passes with 254 tests and 1 optional browser skip; Ruff, diff checks, and source/wheel builds also pass.