• fish_llr

  • Data (contains all raw data files, with burst level: 4)

    • Quantization (contains all Quantization data files, including Quantization data and labels)
    • Tracking (contains all Tracking data files, including Quantization data and labels)

  • src (contains all source code files)

    • ml_generate_label.py (generate labels for Quantization and Tracking data), control: 0, exp: 1, null: -1
    • ml_Quan_Data_preprocessing.py (extract bout rest/active features with different window size (1min, 2min, 30min post light stimulus), and save data in Processed_data/quantization/batch{}/features folder)
    • ml_classification.py (clustering data with features calculated from Quan_Data_preprocessing.py), results saved in \
    • Analysis_Results/ML_results/{fish_type}/Quan_Data_Classification/acc.csv
    • Visualize.py (visualize data with features calculated from classification results.

  • stats_analysis

    • stats_data_cleaning.py (cleaning burst data, and save data in Processed_data/quantization/Tg/stat_data folder)
    • stats_visualize_burst.R (visualize burst level data, save figures in Figures/Stats/Quantization/Tg)

  • Processed_data (contains all processed data files)

    • quantization (contains all Quantization data files, including Quantization data and labels)
    • tracking (contains all Tracking data files, including Quantization data and labels)

1. generate label with ml_generate_label.py and save in Data/Quantization/{fish_type}/{}W-batch{}/{}.npy';
2. get burst duration from data and check if fish do not move all the time, by stats_data_cleaning_burst.py
3. Generate background image into 'Figures/background_image/{}W-batch{}/{}W-60h-{}dpf-01', example png name is 'background_xxxx_A1.png', where xxxx is the frame number, A1 is the well name;
4. Extract light intensity with video_extract.py, save figures to 'Figures/light_intensity/Tg/ {}W-batch{}/{}W-60h-{}dpf-01/{}_{}raw.png' & 'Figures/light_intensity/Tg/{}W-batch{}/ {}W-60h-{}dpf-01/{}{}rm_baseline.png'; save processed_data to 'Processed_data/light_intensity/Tg/{}W-batch{}/ {}W-60h-{}dpf-01/{}W-60h-{}dpf-01/{}{}.pickle';

1. (New) Normalized data, and Extract bout rest/active features (within 30min after stimulus) with ml_normalized_activity.py,
2. (Old) extract bout rest/active features (within 30min after stimulus) with ml_Quan_Data_preprocessing.py, results in Processed_data/quantization/Tg/batch{}/features/{}W-60h-{}dpf-0{}-30-min.csv;
3. do classification with ml_classification.py, result in Analysis_Results/ML_results/Tg/Quan_Data_Classification/ feature_selection/{}W/acc-batch{}.csv;
4. Visualize classification results with ml_acc_visualize_transgenic.py;

1. visualize and statistical comparison for acute response(stats_normalized_by_second.py), collect short-time response, normalized using batch, baseline activity, and light-sensitive level, and compare the differences using hotelling t2 test (stats_burst_duration.py),
2. visualize and statistical comparison for 30 minutes activity (stats_normalized_by_min.py), collect response, normalized using batch, baseline activity, and light-sensitive level, and compare the differences using hotelling t2 test (stats_burst_duration.py),

Statistics for swimming distance

1. generate label with ml_generate_label.py and save in Data/Quantization/Tg/{}W-batch{}/{}.npy
2. get swimming distance (sum) from data and summarise data into one file, by stats_data_cleaning_distance.py, results in Processed_data/tracking/Tg/stat_data named as all_1.2w_60h_batch1_burst4.csv
3. Visualize swimming distance with stats_visualize_distance.R, result in Figures/Stats/Tracking/Tg/all/scale/batch{}, all means all activity (mid + large), raw is raw, scale is min_max scaling.
4. Visualize batches with stats_visualize_distance_batch.R, result in Figures/Stats/Tracking/Tg/all/scale/batches, all means all activity (mid + large), raw is raw, scale is min_max scaling.