#9780 - Save the polymer type switcher (RNA/DNA/PEP) to browser cache

[mariam-khutuashvili]

How the feature works? / How did you fix the issue?

(Screenshots, videos, or GIFs, if applicable)

Description

The Macromolecules polymer-type switcher (RNA / DNA / PEP) now remembers the last-selected type in the browser
cache (localStorage) and restores it when Macromolecules mode is opened, instead of always resetting to RNA.
Defaults to RNA when nothing is stored.

Requirements covered

• Saves the last switcher option to browser cache, persisted on every change path: clicking the switcher and the
  Ctrl+Alt+R/D/P hotkeys (both funnel through the single changeSequenceTypeEnterMode event).
• On opening Macromolecules mode, the polymer type is restored from cache.
• Defaults to RNA when no value is stored, or when the stored value is invalid.
• The restored type also drives the default monomer-library tab (PEP → Peptides tab), including in snake/flex modes
  where the switcher is not visible — the switcher component is always mounted, so the restore-on-open dispatch
  flows through the existing type→tab mapping.

Changes

• constants.ts — added SEQUENCE_TYPE_STORAGE_KEY.
• helpers/sequenceTypeStorage.ts (new) — getPersistedSequenceType() (reads and validates against SequenceType,
  falls back to RNA) and persistSequenceType(), following the existing localStorageWrapper convention.
• helpers/sequenceTypeStorage.test.ts (new) — unit tests: default RNA, persist/restore round-trip, invalid value →
  RNA.
• SequenceTypeGroupButton.tsx — persist the type on change, and restore from cache on mount instead of hardcoding
  RNA.

Check list

• unit-tests written
• e2e-tests written
• documentation updated
• PR name follows the pattern #1234 – issue name
• branch name doesn't contain '#'
• PR is linked with the issue
• base branch (master or release/xx) is correct
• task status changed to "Code review"
• reviewers are notified about the pull request

[claude]

Review summary

Small, focused change with a clean helper and good test coverage of the validation path. A few things worth addressing before merge — all noted as inline comments:

• Handler leak (major). The new persistSequenceType call lives inside a changeSequenceTypeEnterMode listener that the effect cleanup doesn't remove. Pre-existing, but amplified by the new side effect — easy to fix in this PR.
• Unguarded JSON.parse in the helper (minor). localStorageWrapper.getItem will throw on a malformed stored value and crash the mount effect before isSequenceType runs. Worth a try/catch and a test for the malformed-JSON case.
• Wording nit. "browser cache" in the inline comment → "localStorage" to match the surrounding codebase.

Security review found nothing of concern (strict enum equality on the deserialized value, no DOM/HTML/navigation sink). No public-API or changelog doc updates are missing — the analogous PRESET_PHOSPHATE_FILTER_STORAGE_KEY feature shipped the same way.