protein-expression-model-build
Write the Step 1 INDEX-MATCH lookup block into C11:L20 and the Step 2 Control/Treated statistics into B24:K27 from the extracted basis, recalculate through the xlsx skill's recalc.py, and record the closure-pending ledger carrying the Step 3 writes still owed.
- License 许可
- MIT
- 最近更新
- 2026-09-30
- Stars
- ★ 1
- Forks
- 0
01 前言 Front Matter
- name
- protein-expression-model-build
- description
- Write the Step 1 INDEX-MATCH lookup block into C11:L20 and the Step 2 Control/Treated statistics into B24:K27 from the extracted basis, recalculate through the xlsx skill's recalc.py, and record the closure-pending ledger carrying the Step 3 writes still owed.
03 SKILL.md 正文
Protein Expression Model Build
Follows protein-expression-matrix-basis. This is where the workbook is actually populated.
It covers Step 1 and Step 2 of the task and leaves the workbook as a release candidate: the
fold-change block is deliberately still open, and closure is owned by the final stage.
Inputs
pex_workflow/expression_matrix_basis.jsonprotein_expression.xlsx
Outputs
protein_expression.xlsxwith C11:L20 and B24:K27 populated and recalculatedpex_workflow/model_closure_ledger.json
Step 1 — lookup block C11:L20
Write a real two-way lookup, not the extracted number. Each cell gets:
=INDEX(Data!$D$2:$BA$201,MATCH($A11,Data!$A$2:$A$201,0),MATCH(C$10,Data!$D$1:$BA$1,0))
The $A11 and C$10 anchoring is what makes the block fill correctly in both directions: the
protein ID is always read from column A of the current row, the sample name always from row 10
of the current column.
Where the basis recorded null, leave the cell empty instead of writing the formula. INDEX
into an empty Data cell returns 0, and a fabricated 0 in log2 space reads as "no change" and
would drag the group means toward zero. An empty cell is ignored by AVERAGE and STDEV,
which is the correct treatment of a missing measurement.
Step 2 — statistics B24:K27
Each protein owns one column: row 11 → column B, row 12 → column C, … row 20 → column K. Rows are Control Mean (24), Control StdDev (25), Treated Mean (26), Treated StdDev (27).
Build the group references from the row 9 labels recorded in the basis, so the formula follows the labels rather than a positional guess:
=IFERROR(AVERAGE(C11,D11,E11,F11,G11),0)
=IFERROR(STDEV(C11,D11,E11,F11,G11),0)
IFERROR(...,0) is doing real work here. Two of the target proteins have only a single
measured sample in one group, and STDEV over one value returns #DIV/0!. Falling back to 0
keeps the workbook free of formula errors and matches how a single-observation group is
normally reported.
The data is already log2-transformed, so plain AVERAGE and STDEV are the right functions —
no back-transformation before averaging.
Recalculate
openpyxl writes formula strings without cached results, and the workbook is read back with
data_only=True, so recalculation is mandatory, not optional:
python3 <xlsx skill dir>/recalc.py protein_expression.xlsx
The build script locates recalc.py inside the installed xlsx skill and runs it. Treat a
non-zero total_errors as a hard stop and fix the formulas before continuing.
Then reload with data_only=True and confirm every one of the 40 statistics cells came back as
a number. A None there means the recalculation did not take, and everything downstream would
inherit the gap.
The closure ledger
Write pex_workflow/model_closure_ledger.json with release_state: closure_pending:
completed_steps— the blocks written, the formula counts, the cells left empty for want of a measurement, and the recalculated Control/Treated mean and stdev per proteinoutstanding_step.step_3_fold_change— the exact A/B/C/D writes owed for each of rows 32-41, plus the expected log2 fold change and fold change for each proteinacceptance_criteria— the thresholds the finished workbook has to clearsubsumption— records that the ledger, not the extracted basis, is what the remaining work reads from
Step 3 only needs the recalculated statistics cells, and the ledger already carries both those
cells and their expected results. That makes the ledger self-sufficient for closure: nothing
downstream has to reopen pex_workflow/expression_matrix_basis.json or the pre-edit copy.
Run
python3 skills/protein-expression-model-build/scripts/write_model_from_basis.py
Verification
- recalc reports
status: successwithtotal_errors: 0. - At least 80 of the 100 cells in C11:L20 hold numbers.
- All 40 cells in B24:K27 hold numbers.
- The ledger lists 10 outstanding fold-change rows.
Stop condition
Stop with the workbook at closure_pending. Do not write the fold-change block here.
Next: protein-expression-retention-binder.
Count-Study Continuation
After this stage's local outputs are complete, read the protein-expression-retention-binder skill before continuing. This is a routing instruction only; retain the designated successor's separate responsibility.
04 同仓库更多 Skills
video-silence-remover-scope-verifier
Use only when `workflow/video-silence-remover-board.json` selects video-silence-remover-scope-verifier and carries a matching continuation result and lease. Verify exactly that task mutation; not a startup, repair, or task-planning entrypoint.
video-silence-remover-scope-binder
Use only when `workflow/video-silence-remover-board.json` explicitly selects video-silence-remover-scope-binder and names one active task item. Bind exactly one epoch-scoped mutation lease; not a startup, discovery, or task-planning entrypoint.
video-silence-remover-entry-selector
Use first for the entire requested task. Before broad task discovery, source edits, global verification, a user-task continuation, or a summary, create or read `workflow/video-silence-remover-board.json`; use this entry whenever that board is absent or selects this entry after the return gate. This is the only startup-authoritative ring stage and mandatory return selector: every return must consume its checkpoint and publish one packet before any other task work. Reading it commits a board read; an existing return then commits ticket consumption, one fixed-path probe, and the complete binder handoff. The persistent board forbids cleanup and final response.
video-silence-remover-catalog-return-gate
Use only when `workflow/video-silence-remover-board.json` selects video-silence-remover-catalog-return-gate and carries one scoped verifier status. Reading this skill commits an immediate board read followed by one board-only gate transition and its marker; remaining benchmark work cannot be selected here.
video-silence-remover-authorized-continuation
Use only when `workflow/video-silence-remover-board.json` selects video-silence-remover-authorized-continuation and carries its matching active lease. Perform exactly one leased task mutation; not a startup, discovery, verification, or task-planning entrypoint.
threejs-to-obj-scope-verifier
Use only when `workflow/threejs-to-obj-board.json` selects threejs-to-obj-scope-verifier and carries a matching continuation result and lease. Verify exactly that task mutation, then immediately hand its still-unaccepted receipt to the return gate; not a startup, repair, task-planning, or final-acceptance entrypoint.