Code base for all-active model generation, evaluation and analysis (main engine: bluepyopt).

• Method developed and present in Nandi et al. (2020): https://www.biorxiv.org/content/10.1101/2020.04.09.030239v1

• Main developers: Ani Nandi, Werner van Geit, Tom Chartrand, Anatoly Buchin and Costas A. Anastassiou

• Genetic algorithm in action: selection + evaluation + evolution

• Python 3.5+
• Neuron 7.5 (compiled with Python support)
• BluePyOpt
• eFEL
• AllenSDK
• NeuroM
• Pandas
• Seaborn

pip install git+https://github.com/AllenInstitute/All-active-Workflow # install directly from the repository

or

git clone https://github.com/AllenInstitute/All-active-Workflow # clone repository from githubcd All-active-Workflow pip install -e .# install in editable mode

In a Unix environment :

$ source activate conda-env # conda environment with all dependencies $ launch_optimjob --help # Look at the options $ launch_optimjob --input_json job_config.json

• Configurable staged optimization

  job_config.json

  {"cty_config":{"cell_id": "483101699" }, "job_config":{"highlevel_jobconfig":{"conda_env": "ateam_opt", "axon_type": "stub_axon", "data_source": "web", "ephys_dir": "ephys_data", "non_standard_nwb": false, "feature_stimtypes": [ "Long Square" ], "feature_names_path": "feature_set_all.json", "compiled_modfiles_dir": "x86_64", "job_dir": "483101699_benchmark_timeout" }, "stage_jobconfig": [{"stage_name": "Stage0", "stage_stimtypes": [ "Long Square" ], "stage_features": "feature_set_stage0.json", "stage_parameters": "param_bounds_stage0.json", "filter_rule": "filter_feat_proto_passive", "offspring_size": 512, "max_ngen": 50, "optim_config":{"nengines": 256, "nnodes": 16, "qos": "celltypes", "nprocs": 16, "error_stream": "job.err", "output_stream": "job.out", "jobmem": "100g", "jobtime": "5:00:00", "ipyparallel": true, "ipyparallel_db": "sqlitedb", "main_script": "Optim_Main.py" }, "analysis_config":{"main_script": "analyze_stagejob.py" }, "seed": [ 1 ] },{"stage_name": "Stage1", "stage_stimtypes": [ "Long Square" ], "stage_features": "feature_set_stage1.json", "stage_parameters": "param_bounds_stage1.json", "filter_rule": "filter_feat_proto_passive", "offspring_size": 512, "max_ngen": 50, "optim_config":{"nengines": 256, "nnodes": 16, "qos": "celltypes", "nprocs": 16, "error_stream": "job.err", "output_stream": "job.out", "jobmem": "100g", "jobtime": "5:00:00", "ipyparallel": true, "ipyparallel_db": "sqlitedb", "main_script": "Optim_Main.py" }, "analysis_config":{"main_script": "analyze_stagejob.py" }, "seed": [ 1 ] },{"stage_name": "Stage2", "stage_stimtypes": [ "Long Square" ], "stage_features": "feature_set_stage2.json", "stage_parameters": "param_bounds_stage2_mouse_spiny.json", "filter_rule": "filter_feat_proto_active", "AP_initiation_zone": "axon", "offspring_size": 512, "cp_backup_dir": "checkpoints_backup", "max_ngen": 200, "optim_config":{"nengines": 256, "nnodes": 16, "qos": "celltypes", "nprocs": 16, "error_stream": "job.err", "output_stream": "job.out", "jobmem": "150g", "jobtime": "12:00:00", "ipyparallel": true, "ipyparallel_db": "sqlitedb", "main_script": "Optim_Main.py" }, "analysis_config":{"main_script": "analyze_stagejob.py", "ipyparallel": true, "ipyparallel_db": "nodb", "error_stream": "analysis.err", "output_stream": "analysis.out", "nengines": 40, "nnodes": 4, "nprocs": 10, "jobtime": "10:00:00", "jobmem": "100g", "qos": "celltypes" }, "seed": [ 1, 2, 3, 4 ], "run_hof_analysis": true, "run_peri_comparison": false, "depol_block_check": true, "add_fi_kink": true, "calc_model_perf": true, "model_postprocess": true, "calc_time_statistics": true, "timeout": 300, "hoc_export": true } ] } } 

Instructions are included to run these compute intensive jobs on 4 different architectures (Allen Institute HPC cluster, NERSC, BBP5 and AWS).

We are planning on occasional updating this tool with no fixed schedule. Community involvement is encouraged through both issues and pull requests.