v2ecoli ParCa Workflow Report

Overview

Generated 2026-04-16 08:16:55

v2parca checkpoints: /Users/eranagmon/code/v2ecoli-merge-parca/out/sim_data. vEcoli intermediates: out/original_intermediates.

Per-step runtime

Step-by-step availability

This Run

Summary of the most recent parca_bigraph.py invocation (checkpoints + runtimes from /Users/eranagmon/code/v2ecoli-merge-parca/out/sim_data).

0. EcoCyc API

Ten TSVs in reconstruction/ecoli/flat/ are refreshable from BioCyc via https://websvc.biocyc.org/wc-get?type=<file_id>. These are the upstream source of truth for genes, RNAs, proteins, reactions, DNA binding sites, and equilibrium/complexation reactions. The other 120-odd files under flat/ are either curated by the Covert lab or derived from other databases (EcoCyc + ModelSEED + literature). Rerun the comparison report with --fetch-biocyc to refresh these files in-place.

Last fetched: 2026-04-15 23:00:06 — 10/10 files OK.

Raw Data — flat-file knowledge base

Loaded by KnowledgeBaseEcoli at composite-construction time and passed into InitializeStep's config (not a port, which is why it does not appear on step 1's declared data flow table). Total: 133 files (10 EcoCyc-refreshable, 123 curated/derived), 12.9 MB. Source: v2ecoli / processes / parca / reconstruction / ecoli / flat ↗.

Knowledge base statistics

Step 1 — initialize

Initialize sim_data + scatter

Bootstraps SimulationDataEcoli from the flat-file knowledge base. Loads TSVs into typed tables (genes, RNAs, proteins, reactions, metabolite concentrations, genome sequence) and scatters them across the 20 subsystem objects (Mass, Constants, Transcription, Translation, Metabolism, …) that the rest of the pipeline reads from.

Runtime

Declared data flow

Each Step declares INPUT_PORTS (stores it reads) and OUTPUT_PORTS (stores it writes). Port names are resolved to absolute store paths via STORE_PATH in v2parca/composite.py, and the composite wires each port to the corresponding location in the nested bigraph. The role column shows the description given in the step module (e.g. sim_data.transcription means "mirrors the transcription subsystem on SimulationDataEcoli"; overwrite means the port replaces whatever value lives at that store).

Step purpose: Step 1 — initialize (scatter). Bootstrap the pipeline from raw KB data.

Inputs — reads (0)

(none — this step has no upstream subsystem reads; it only consumes ordering tokens: —)

Outputs — writes (32)

Plus ordering tokens: tick_1 — these enforce Step execution order in the bigraph but carry no data.

Scalar state

Distributions

RNA expression — basal

RNA synthesis prob — basal

RNA deg rates

Cistron deg rates

Protein deg rates

Translation efficiencies

Km endoRNase (transcribed)

Km endoRNase (mature)

Distribution numerical summary

Step 2 — input_adjustments

Pre-fitted adjustments to expression + deg rates

Applies hand-curated corrections before any fitting: edits to RNA expression (rRNA + mRNA), protein and RNA degradation rates, translation efficiencies, and the small-molecule concentrations that the fitting loop uses as targets. Many of these are Keasling-lab / Covert-lab adjustments that reflect empirical observations missing from pure bioinformatic sources.

Runtime

Declared data flow

Each Step declares INPUT_PORTS (stores it reads) and OUTPUT_PORTS (stores it writes). Port names are resolved to absolute store paths via STORE_PATH in v2parca/composite.py, and the composite wires each port to the corresponding location in the nested bigraph. The role column shows the description given in the step module (e.g. sim_data.transcription means "mirrors the transcription subsystem on SimulationDataEcoli"; overwrite means the port replaces whatever value lives at that store).

Step purpose: Step 2 — input_adjustments. Apply literature-curated overrides before fitting.

Inputs — reads (4)

Plus ordering tokens: tick_1 — these enforce Step execution order in the bigraph but carry no data.

Outputs — writes (3)

Plus ordering tokens: tick_2 — these enforce Step execution order in the bigraph but carry no data.

Scalar state

Distributions

RNA expression — basal

RNA synthesis prob — basal

RNA deg rates

Cistron deg rates

Protein deg rates

Translation efficiencies

Km endoRNase (transcribed)

Km endoRNase (mature)

Distribution numerical summary

Step 3 — basal_specs

Build basal cell specifications

Builds the basal cell specification — expression levels, synthesis probabilities, bulk-molecule counts, and doubling time for a reference growth condition (M9 + glucose, no TF perturbations). This is the spec every later condition perturbs away from.

Runtime

Declared data flow

Each Step declares INPUT_PORTS (stores it reads) and OUTPUT_PORTS (stores it writes). Port names are resolved to absolute store paths via STORE_PATH in v2parca/composite.py, and the composite wires each port to the corresponding location in the nested bigraph. The role column shows the description given in the step module (e.g. sim_data.transcription means "mirrors the transcription subsystem on SimulationDataEcoli"; overwrite means the port replaces whatever value lives at that store).

Step purpose: Step 3 — basal_specs. Fit the minimal-medium (basal) condition.

Inputs — reads (17)

Plus ordering tokens: tick_2 — these enforce Step execution order in the bigraph but carry no data.

Outputs — writes (5)

Plus ordering tokens: tick_3 — these enforce Step execution order in the bigraph but carry no data.

Scalar state

Distributions

RNA expression — basal

RNA synthesis prob — basal

RNA deg rates

Cistron deg rates

Protein deg rates

Translation efficiencies

Km endoRNase (transcribed)

Km endoRNase (mature)

Distribution numerical summary

cell_specs (per-condition max rel Δ)

Step 4 — tf_condition_specs

Per-TF + combined condition cell specs

Replays step 3 once per transcription-factor activation state, producing a cell spec for each (condition, TF) pair. The bulk of the wall time (~50 s even in fast mode) goes into parallel evaluation of the TF conditions required to anchor the binding-probability fit in step 6.

Runtime

Declared data flow

Each Step declares INPUT_PORTS (stores it reads) and OUTPUT_PORTS (stores it writes). Port names are resolved to absolute store paths via STORE_PATH in v2parca/composite.py, and the composite wires each port to the corresponding location in the nested bigraph. The role column shows the description given in the step module (e.g. sim_data.transcription means "mirrors the transcription subsystem on SimulationDataEcoli"; overwrite means the port replaces whatever value lives at that store).

Step purpose: Step 4 — tf_condition_specs. Repeat the basal fit under each TF condition.

Inputs — reads (25)

Plus ordering tokens: tick_3 — these enforce Step execution order in the bigraph but carry no data.

Outputs — writes (2)

Plus ordering tokens: tick_4 — these enforce Step execution order in the bigraph but carry no data.

Scalar state

Distributions

RNA expression — basal

RNA synthesis prob — basal

RNA deg rates

Cistron deg rates

Protein deg rates

Translation efficiencies

Km endoRNase (transcribed)

Km endoRNase (mature)

Distribution numerical summary

cell_specs (per-condition max rel Δ)

Step 5 — fit_condition

Bulk distributions + translation supply rates

Solves for the bulk-molecule distribution and translation supply rates that make the basal-plus-condition cell specs self-consistent. Runs a fixed-point iteration inside CVXPY; ~70 min in fast mode — the single most expensive step and the reason checkpointing was added.

Runtime

Declared data flow

Each Step declares INPUT_PORTS (stores it reads) and OUTPUT_PORTS (stores it writes). Port names are resolved to absolute store paths via STORE_PATH in v2parca/composite.py, and the composite wires each port to the corresponding location in the nested bigraph. The role column shows the description given in the step module (e.g. sim_data.transcription means "mirrors the transcription subsystem on SimulationDataEcoli"; overwrite means the port replaces whatever value lives at that store).

Step purpose: Step 5 — fit_condition. Bulk-molecule distributions + translation supply.

Inputs — reads (16)

Plus ordering tokens: tick_4 — these enforce Step execution order in the bigraph but carry no data.

Outputs — writes (2)

Plus ordering tokens: tick_5 — these enforce Step execution order in the bigraph but carry no data.

Step 6 — promoter_binding

TF-promoter binding probabilities (CVXPY)

Fits TF-promoter binding probabilities. For each (RNA, condition) pair, solves a small convex program so that synth_prob = basal + Σ r · P reproduces the synthesis probability step 4 computed, with non-negative recruitment strengths r and equilibrium-derived binding probabilities P.

Runtime

Declared data flow

Each Step declares INPUT_PORTS (stores it reads) and OUTPUT_PORTS (stores it writes). Port names are resolved to absolute store paths via STORE_PATH in v2parca/composite.py, and the composite wires each port to the corresponding location in the nested bigraph. The role column shows the description given in the step module (e.g. sim_data.transcription means "mirrors the transcription subsystem on SimulationDataEcoli"; overwrite means the port replaces whatever value lives at that store).

Step purpose: Step 6 — promoter_binding. Fit TF binding probabilities and recruitment.

Inputs — reads (21)

Plus ordering tokens: tick_5 — these enforce Step execution order in the bigraph but carry no data.

Outputs — writes (4)

Plus ordering tokens: tick_6 — these enforce Step execution order in the bigraph but carry no data.

Step 7 — adjust_promoters

Ligand concentrations + RNAP recruitment

Adjusts ligand concentrations and RNAP recruitment strengths to account for the TF-promoter fit. Produces the final per-condition RNA synthesis probabilities that the online model will use.

Runtime

Declared data flow

Each Step declares INPUT_PORTS (stores it reads) and OUTPUT_PORTS (stores it writes). Port names are resolved to absolute store paths via STORE_PATH in v2parca/composite.py, and the composite wires each port to the corresponding location in the nested bigraph. The role column shows the description given in the step module (e.g. sim_data.transcription means "mirrors the transcription subsystem on SimulationDataEcoli"; overwrite means the port replaces whatever value lives at that store).

Step purpose: Step 7 — adjust_promoters. Back-solve ligand levels + RNAP recruitment.

Inputs — reads (23)

Plus ordering tokens: tick_6 — these enforce Step execution order in the bigraph but carry no data.

Outputs — writes (3)

Plus ordering tokens: tick_7 — these enforce Step execution order in the bigraph but carry no data.

Step 8 — set_conditions

Per-nutrient dicts + mass rescaling

Materializes everything into the per-nutrient dictionaries the online simulation indexes by condition: external concentrations, doubling times, rescaled masses. Thin wrapper over the step 3-7 outputs.

Runtime

Declared data flow

Each Step declares INPUT_PORTS (stores it reads) and OUTPUT_PORTS (stores it writes). Port names are resolved to absolute store paths via STORE_PATH in v2parca/composite.py, and the composite wires each port to the corresponding location in the nested bigraph. The role column shows the description given in the step module (e.g. sim_data.transcription means "mirrors the transcription subsystem on SimulationDataEcoli"; overwrite means the port replaces whatever value lives at that store).

Step purpose: Step 8 — set_conditions. Build per-condition mass + expression tables.

Inputs — reads (11)

Plus ordering tokens: tick_7 — these enforce Step execution order in the bigraph but carry no data.

Outputs — writes (4)

Plus ordering tokens: tick_8 — these enforce Step execution order in the bigraph but carry no data.

Step 9 — final_adjustments

ppGpp kinetics + amino-acid supply constants

Last-mile adjustments: ppGpp kinetics (for the stringent response), amino-acid supply constants, and mechanistic kcat/KM corrections for amino-acid export, uptake, and supply. Some of these fail in fast mode because the reduced TF-condition set leaves key metabolites unconstrained — those are caught and logged as warnings so the pipeline still produces a comparable pickle.

Runtime

Declared data flow

Each Step declares INPUT_PORTS (stores it reads) and OUTPUT_PORTS (stores it writes). Port names are resolved to absolute store paths via STORE_PATH in v2parca/composite.py, and the composite wires each port to the corresponding location in the nested bigraph. The role column shows the description given in the step module (e.g. sim_data.transcription means "mirrors the transcription subsystem on SimulationDataEcoli"; overwrite means the port replaces whatever value lives at that store).

Step purpose: Step 9 — final_adjustments. Kinetic constants the online model needs.

Inputs — reads (29)

Plus ordering tokens: tick_8 — these enforce Step execution order in the bigraph but carry no data.

Outputs — writes (3)

Plus ordering tokens: tick_9 — these enforce Step execution order in the bigraph but carry no data.

Composition Diagram