DB is organised into tables:
- MODEL: model names and specific parameters
- EXPERIMENT: location, description and parameters of data files
- RUN: priors, model and data pertaining to a set of MCMC runs
- TASK: MCMC run on a cell, RUNS are subdivided into TASKS
- MICROSCOPE: microscope names
- USER: user names
- LOG: log output from MCMC runs
Possible future tables:
- CELL_LINE: cell line descriptions, e.g. HeLa-J WT
- CONDITION: treatments, e.g. nocadozale x mM
-
As always, try 'help command' for more info.
-
Before any use must execute:
- dbInit
-
Query database with:
- dbShowUsers
- dbShowMicroscopes
- dbShowModels
- dbShowExperiments
- dbShowTasks
-
To import data already in trajData format, use:
- dbAddExperiment(initials,microscope,file,date,cell_line,name,desc);
-
To import data in KiT format, use:
- dbImportKitData(directory);
- Several optional arguments are possible, if not supplied they will be queried
-
Setup a set of MCMC runs with:
- dbSetupRun('name','myrun1');
- All parameters are optional and will be queried if needed, except name.
- Priors can be specified as a struct or a string, e.g.:
- p.L = [10 0.5]; p.kappa = [0.5 1]; ...., or
- p = 'L=[10 0.5];kappa=[0.5 1];'
-
Start executing runs with:
- dbExecute for multi-core serial execution, or
- dbExecute('batch') for single-core per task parallel batch execution.
-
Check convergence diagnostics:
- convReport = dbConvergence(run_id);
-
Rerun MCMCs to improve convergence with:
- dbRerun(run_id);
-
To compute marginal likelihoods for a run:
- dbChen(run_id);
- dbChib(run_id);