Skip to content

nerd probe_timecourse

nerd probe_timecourse fits chemical probing time-course data stored in probe_fmod_values, producing per-nucleotide parameters (free fits, global deg fits, constrained refits). Results are written to probe_tc_fit_runs/probe_tc_fit_params and JSON artifacts for each reaction group.

nerd probe_timecourse --config PATH/TO/config.yaml --db PATH/TO/nerd.sqlite

Omitting --db defaults to <run.output_dir>/nerd.sqlite.

Configuration Layout

run:
  label: probe_tc_demo
  output_dir: examples

probe_timecourse:
  engine: python_baseline
  rounds:
    - round1_free
    - round2_global
    - round3_constrained
  rg_ids: [101, 102]
  valtype:
    - modrate
    - GAmodrate
  min_points: 3
  overwrite: true

  engine_options:
    global_selection: ac_only
    global_filters:
      r2_threshold: 0.75
    initial_kdeg: 0.0025

Key Fields

Field Description
engine Registered timecourse engine (python_baseline is the default).
rounds Subset of the three-stage workflow (free, global, constrained).
rg_ids Reaction-group IDs to process; leave empty to process all.
valtype Which value type(s) to use. Provide a string for one, or a list (e.g., [modrate, GAmodrate]) to aggregate multiple.
min_points Minimum points required per nucleotide (default 3).
overwrite When true, existing fits in probe_tc_fit_runs are replaced.
engine_options Passed to the engine; e.g., global selection filters, initial guesses.

Additional filters: nt_ids, species, buffer, construct, fit_kind, etc. (ignored unless supported by the engine). See the example configs under examples/probing_timecourse_fits/.

Workflow Overview

  1. Round 1 (round1_free) – independent per-nucleotide fits (k_obs, k_deg, f_mod0).
  2. Round 2 (round2_global) – global k_deg using filtered nucleotides (A/C only, R² threshold, etc.).
  3. Round 3 (round3_constrained) – refit each nucleotide with k_deg fixed from round 2.

You can run any subset of rounds depending on the goal (e.g., skip global and only produce free fits by listing round1_free).

Outputs

  • Per-nucleotide parameters and diagnostics in probe_tc_fit_params (tall format).
  • Run-level metadata in probe_tc_fit_runs (fit kind, rg_id, nt_id, timestamp).
  • JSON summaries under <output_dir>/<label>/probe_timecourse_latest/results/.

Examples

# Full three-round fit for selected reaction groups
nerd probe_timecourse --config examples/probing_timecourse_fits/config.yaml                       --db examples/nerd.sqlite

# Only run free fits
nerd probe_timecourse --config configs/probe_tc_free.yaml --db nerd.sqlite

# Constrained refits using existing global kdeg
nerd probe_timecourse --config configs/probe_tc_constrained.yaml --db nerd.sqlite

Once fits are completed, downstream tasks (e.g., tempgrad_fit with data_source: probe_tc) can reuse the stored kinetics for Arrhenius analysis.