Pipeline

What turns inputs into outputs

The functions, by the names the code gives them. Each one reads inputs someone wrote and writes outputs nobody edits; each page says what it reads, what it writes, where its code is, and every element it produced. The invariant in one view: agents write source, the pipeline regenerates, humans approve.

End to end

Inputs

Datawhat was measured
Metadatawhat was declared
Textwhat was written

Function

prepare_data()R function
build_ard()R function
render_display()R function
render_rtf()R function
regenerate()R function
regenerate_values()R function
assemble.mjsNode script
site.mjsNode script

Outputs

DisplaysARDs and rendered tables
Valuesthe named-number store
Documentsthe assembled reports

prepare_data() R function

Reads each dataset the specs name from its source lane, applies the documented derivations, and hands the prepared frames to the ARD builder; every ARD records each frame's hash and row count.

Inputs

vendored transport filesthe study's own package
the source registrydata_sources(): which lane each dataset resolves to
the study modelanalysis-set flags, arm labels

Function

prepare_data()pipeline/R/data-prep.R

Outputs

prepared datasetsin memory; hash and rows recorded in every ARD envelope

Code: pipeline/R/data-prep.R

Produced

adsl Subject-level analysis dataset · read by 30 displays
adae Adverse events · read by 5 displays
advs Vital signs · read by 3 displays
adqsadas ADAS-Cog (11) · read by 8 displays
adqscibc CIBIC+ · read by 4 displays
adqsnpix NPI-X · read by 1 display
adtte Time to first dermatologic event · read by 1 display
adcm Concomitant medications (analysis dataset) · read by 1 display
dm Demographics (SDTM domain) · read by 1 display

build_ard() R function

Runs each analysis in analysis.yaml — built-in or the display's own custom.R — over the prepared datasets and stacks the results into the ARD, one row per statistic, with the population record.

Inputs

analysis.yamlwhat to compute
custom.Rwhere a built-in method does not fit
prepared datasets
the study modelarms in print order

Function

build_ard()pipeline/R/ard.R

Outputs

ARD rowsopencsr/ard/v1, plus the population record

Code: pipeline/R/ard.R

Produced

t-demographics Demographic and Baseline Characteristics · v007
t-disposition Subject Disposition · v003
t-populations Summary of Populations · v003
t-end-of-study Summary of End of Study Data · v003
f-derm-time-to-event Time to Dermatologic Event by Treatment Group · v002
t-cibic-week24 Primary Endpoint Analysis: CIBIC+ — Summary at Week 24 — LOCF · v004
t-cibic-week8 CIBIC+ — Summary at Week 8 — LOCF · v002
t-cibic-week16 CIBIC+ — Summary at Week 16 — LOCF · v002
t-cibic-categorical CIBIC+ — Categorical Analysis — LOCF · v002
t-eff-adas-wk24 Primary Endpoint Analysis: ADAS-Cog (11) - Change from Baseline to Week 24 - LOCF · v002
t-eff-adas-wk8 ADAS-Cog (11) - Change from Baseline to Week 8 - LOCF · v002
t-eff-adas-wk16 ADAS-Cog (11) - Change from Baseline to Week 16 - LOCF · v002
t-eff-adas-wk24-completers ADAS-Cog (11) - Change from Baseline to Week 24 - Completers at Week 24, Observed Cases, Windowed · v002
t-eff-adas-wk24-male ADAS-Cog (11) - Change from Baseline to Week 24 in Male Subjects - LOCF · v002
t-eff-adas-wk24-female ADAS-Cog (11) - Change from Baseline to Week 24 in Female Subjects - LOCF · v002
t-eff-adas-overtime ADAS-Cog (11) - Mean and Mean Change from Baseline over Time · v002
t-eff-adas-mmrm ADAS-Cog (11) - Repeated Measures Analysis of Change from Baseline to Week 24 · v004
t-eff-npix-mean Mean NPI-X Total Score from Week 4 through Week 24 - Windowed · v003
t-ae-overview Overview of Treatment-Emergent Adverse Events · v003
t-ae-common Treatment-Emergent Adverse Events by System Organ Class and Preferred Term · v004
t-exposure Extent of Exposure to Study Drug · v004
l-ae-serious Listing of Serious Adverse Events · v003
t-conmeds Summary of Concomitant Medications · v003
t-vitals Summary of Vital Signs at Baseline and End of Treatment · v002
t-vitals-change Summary of Vital Signs Change from Baseline at End of Treatment · v002
t-weight Summary of Weight and Weight Change from Baseline at End of Treatment · v004
t-ae-incidence Incidence of Treatment Emergent Adverse Events by Treatment Group · v007
t-sae-incidence Incidence of Treatment Emergent Serious Adverse Events by Treatment Group · v006
t-subjects-by-site Summary of Number of Subjects By Site · v002
f-disposition Subject Disposition · v001

render_display() R function

Draws each variant of the display from the ARD and display.yaml: columns, rows, formats, footnotes, and the figure where the display is one.

Inputs

ARD rows
display.yamlhow to show it

Function

render_display()pipeline/R/render.R

Outputs

table.html per variantthe rendered cells

Code: pipeline/R/render.R

Produced

t-demographics Demographic and Baseline Characteristics · v007
t-disposition Subject Disposition · v003
t-populations Summary of Populations · v003
t-end-of-study Summary of End of Study Data · v003
f-derm-time-to-event Time to Dermatologic Event by Treatment Group · v002
t-cibic-week24 Primary Endpoint Analysis: CIBIC+ — Summary at Week 24 — LOCF · v004
t-cibic-week8 CIBIC+ — Summary at Week 8 — LOCF · v002
t-cibic-week16 CIBIC+ — Summary at Week 16 — LOCF · v002
t-cibic-categorical CIBIC+ — Categorical Analysis — LOCF · v002
t-eff-adas-wk24 Primary Endpoint Analysis: ADAS-Cog (11) - Change from Baseline to Week 24 - LOCF · v002
t-eff-adas-wk8 ADAS-Cog (11) - Change from Baseline to Week 8 - LOCF · v002
t-eff-adas-wk16 ADAS-Cog (11) - Change from Baseline to Week 16 - LOCF · v002
t-eff-adas-wk24-completers ADAS-Cog (11) - Change from Baseline to Week 24 - Completers at Week 24, Observed Cases, Windowed · v002
t-eff-adas-wk24-male ADAS-Cog (11) - Change from Baseline to Week 24 in Male Subjects - LOCF · v002
t-eff-adas-wk24-female ADAS-Cog (11) - Change from Baseline to Week 24 in Female Subjects - LOCF · v002
t-eff-adas-overtime ADAS-Cog (11) - Mean and Mean Change from Baseline over Time · v002
t-eff-adas-mmrm ADAS-Cog (11) - Repeated Measures Analysis of Change from Baseline to Week 24 · v004
t-eff-npix-mean Mean NPI-X Total Score from Week 4 through Week 24 - Windowed · v003
t-ae-overview Overview of Treatment-Emergent Adverse Events · v003
t-ae-common Treatment-Emergent Adverse Events by System Organ Class and Preferred Term · v004
t-exposure Extent of Exposure to Study Drug · v004
l-ae-serious Listing of Serious Adverse Events · v003
t-conmeds Summary of Concomitant Medications · v003
t-vitals Summary of Vital Signs at Baseline and End of Treatment · v002
t-vitals-change Summary of Vital Signs Change from Baseline at End of Treatment · v002
t-weight Summary of Weight and Weight Change from Baseline at End of Treatment · v004
t-ae-incidence Incidence of Treatment Emergent Adverse Events by Treatment Group · v007
t-sae-incidence Incidence of Treatment Emergent Serious Adverse Events by Treatment Group · v006
t-subjects-by-site Summary of Number of Subjects By Site · v002
f-disposition Subject Disposition · v001

render_rtf() R function

Writes the same rendered cells as submission-format RTF, hash-recorded in the manifest so a hand-edited copy cannot pass as the pipeline's.

Inputs

the rendered cells

Function

render_rtf()pipeline/R/rtf.R

Outputs

table.rtf per variantsha256 in the manifest

Code: pipeline/R/rtf.R

Produced

t-demographics Demographic and Baseline Characteristics · v007
t-disposition Subject Disposition · v003
t-populations Summary of Populations · v003
t-end-of-study Summary of End of Study Data · v003
f-derm-time-to-event Time to Dermatologic Event by Treatment Group · v002
t-cibic-week24 Primary Endpoint Analysis: CIBIC+ — Summary at Week 24 — LOCF · v004
t-cibic-week8 CIBIC+ — Summary at Week 8 — LOCF · v002
t-cibic-week16 CIBIC+ — Summary at Week 16 — LOCF · v002
t-cibic-categorical CIBIC+ — Categorical Analysis — LOCF · v002
t-eff-adas-wk24 Primary Endpoint Analysis: ADAS-Cog (11) - Change from Baseline to Week 24 - LOCF · v002
t-eff-adas-wk8 ADAS-Cog (11) - Change from Baseline to Week 8 - LOCF · v002
t-eff-adas-wk16 ADAS-Cog (11) - Change from Baseline to Week 16 - LOCF · v002
t-eff-adas-wk24-completers ADAS-Cog (11) - Change from Baseline to Week 24 - Completers at Week 24, Observed Cases, Windowed · v002
t-eff-adas-wk24-male ADAS-Cog (11) - Change from Baseline to Week 24 in Male Subjects - LOCF · v002
t-eff-adas-wk24-female ADAS-Cog (11) - Change from Baseline to Week 24 in Female Subjects - LOCF · v002
t-eff-adas-overtime ADAS-Cog (11) - Mean and Mean Change from Baseline over Time · v002
t-eff-adas-mmrm ADAS-Cog (11) - Repeated Measures Analysis of Change from Baseline to Week 24 · v004
t-eff-npix-mean Mean NPI-X Total Score from Week 4 through Week 24 - Windowed · v003
t-ae-overview Overview of Treatment-Emergent Adverse Events · v003
t-ae-common Treatment-Emergent Adverse Events by System Organ Class and Preferred Term · v004
t-exposure Extent of Exposure to Study Drug · v004
l-ae-serious Listing of Serious Adverse Events · v003
t-conmeds Summary of Concomitant Medications · v003
t-vitals Summary of Vital Signs at Baseline and End of Treatment · v002
t-vitals-change Summary of Vital Signs Change from Baseline at End of Treatment · v002
t-weight Summary of Weight and Weight Change from Baseline at End of Treatment · v004
t-ae-incidence Incidence of Treatment Emergent Adverse Events by Treatment Group · v007
t-sae-incidence Incidence of Treatment Emergent Serious Adverse Events by Treatment Group · v006
t-subjects-by-site Summary of Number of Subjects By Site · v002
f-disposition Subject Disposition · v001

regenerate() R function

One display, one rebuild: prepares the data, builds the ARD, renders every variant, and writes a new iteration folder with its manifest — never editing an earlier one.

Inputs

Function

regenerate()pipeline/R/regenerate.R

Outputs

outputs/<slug>/vNNN/spec snapshots, ard.json, table.html, table.rtf, manifest.json
current.jsonwhich iteration is current

Code: pipeline/R/regenerate.R

Calls prepare_data(), build_ard(), render_display(), render_rtf().

Produced

t-demographics Demographic and Baseline Characteristics · v007
t-disposition Subject Disposition · v003
t-populations Summary of Populations · v003
t-end-of-study Summary of End of Study Data · v003
f-derm-time-to-event Time to Dermatologic Event by Treatment Group · v002
t-cibic-week24 Primary Endpoint Analysis: CIBIC+ — Summary at Week 24 — LOCF · v004
t-cibic-week8 CIBIC+ — Summary at Week 8 — LOCF · v002
t-cibic-week16 CIBIC+ — Summary at Week 16 — LOCF · v002
t-cibic-categorical CIBIC+ — Categorical Analysis — LOCF · v002
t-eff-adas-wk24 Primary Endpoint Analysis: ADAS-Cog (11) - Change from Baseline to Week 24 - LOCF · v002
t-eff-adas-wk8 ADAS-Cog (11) - Change from Baseline to Week 8 - LOCF · v002
t-eff-adas-wk16 ADAS-Cog (11) - Change from Baseline to Week 16 - LOCF · v002
t-eff-adas-wk24-completers ADAS-Cog (11) - Change from Baseline to Week 24 - Completers at Week 24, Observed Cases, Windowed · v002
t-eff-adas-wk24-male ADAS-Cog (11) - Change from Baseline to Week 24 in Male Subjects - LOCF · v002
t-eff-adas-wk24-female ADAS-Cog (11) - Change from Baseline to Week 24 in Female Subjects - LOCF · v002
t-eff-adas-overtime ADAS-Cog (11) - Mean and Mean Change from Baseline over Time · v002
t-eff-adas-mmrm ADAS-Cog (11) - Repeated Measures Analysis of Change from Baseline to Week 24 · v004
t-eff-npix-mean Mean NPI-X Total Score from Week 4 through Week 24 - Windowed · v003
t-ae-overview Overview of Treatment-Emergent Adverse Events · v003
t-ae-common Treatment-Emergent Adverse Events by System Organ Class and Preferred Term · v004
t-exposure Extent of Exposure to Study Drug · v004
l-ae-serious Listing of Serious Adverse Events · v003
t-conmeds Summary of Concomitant Medications · v003
t-vitals Summary of Vital Signs at Baseline and End of Treatment · v002
t-vitals-change Summary of Vital Signs Change from Baseline at End of Treatment · v002
t-weight Summary of Weight and Weight Change from Baseline at End of Treatment · v004
t-ae-incidence Incidence of Treatment Emergent Adverse Events by Treatment Group · v007
t-sae-incidence Incidence of Treatment Emergent Serious Adverse Events by Treatment Group · v006
t-subjects-by-site Summary of Number of Subjects By Site · v002
f-disposition Subject Disposition · v001

regenerate_values() R function

Resolves every named value's address against the committed ARDs, evaluates the declared arithmetic, and writes the store with each value's provenance.

Inputs

values.yamlnames, addresses, arithmetic
committed ARDseach display's current iteration

Function

regenerate_values()pipeline/R/values.R

Outputs

outputs/values/values.jsonvalue, formatted value, ARD row and hash

Code: pipeline/R/values.R

Produced

randomised-n Subjects randomised
treated-n Subjects treated (safety analysis set)
completed-n Subjects who completed the study
discontinued-n Subjects who discontinued
completed-pct Percentage of randomised subjects who completed
age-median-total Median age (years), all subjects
female-n-total Female subjects
ae-any-n-total Subjects with at least one adverse event
ae-any-n-placebo Subjects with at least one adverse event (Placebo)
ae-any-n-low Subjects with at least one adverse event (Xanomeline Low Dose)
ae-any-n-high Subjects with at least one adverse event (Xanomeline High Dose)
ae-any-pct-high Percentage of high-dose subjects with at least one adverse event
safety-n-high Subjects in the safety analysis set (Xanomeline High Dose)
ae-any-n-xanomeline Subjects with at least one adverse event, both xanomeline arms
ae-excess-high-vs-placebo Additional subjects with an adverse event, high dose versus placebo

assemble.mjs Node script

Fills a template object with this study's text, displays and values, numbers the displays from assembly order, records what each block and the document were built from, and runs the seven gates.

Inputs

sections.yaml + assembly.yamlthe document model and what this study puts in it
text blocksapproved ones, for generated prose
current ARDs and rendered variants
values.json
the study modelthe treatment-consistency gate

Function

assemble.mjsscripts/assemble.mjs

Outputs

docs/assembled/<document>.jsonthe document as data, with every gate's result and its inputs record
docs/assembled/<document>.htmlthe rendered document

Code: scripts/assemble.mjs

Gates

structure
binding resolution
numeric fidelity
cross-references
values
treatment consistency
display coverage

Produced

Clinical Study Report template ich-e3
Study Synopsis template e3-synopsis
Post-Text Display Package template display-package
Abbreviated Clinical Study Report (ICH E3) template e3-abbreviated

site.mjs Node script

Builds every page of this site from the registry, the outputs, the assembled documents and the evidence — including the pages you are reading — and fails on any internal link that does not resolve or any external resource.

Inputs

site/config.jsonthe registry
outputs/every iteration
docs/assembled/
docs/evidence/ + docs/requirements/

Function

site.mjsscripts/site.mjs

Outputs

site/_build/every page, no external resource

Code: scripts/site.mjs

Produced

Nothing in the site collection yet.