# FRETboard: data and performance evaluation

The simulated and *in vitro* data in this repository were used to evaluate the performance of 
the FRETboard FRET trace analysis tool. Each can be found in their respective directories.
Scripts to rerun the performance evaluation are included as well.

## Contents

Directories `simulated` and `in_vitro` are structured as follows: 
- data
  - `experiment_1`
    - `dats_unlabeled`: plain text files containing traces as accepted by FRETboard
    - `dats_labeled`: plain text files of the traces with additional column with ground truth labels
  - `experiment_2`
    - ...
- eval
  - experiment_1_fb: FRETboard results experiment 1
    - `dat_files`: plain text files with traces and predicted labels
    - `FRETboard_report.html`: the html report as generated by FRETboard
    - `FRETboard_data_transition_rates.csv`: transition rate estimates. This file can be extracted
  - experiment_1_eval: performance evaluation results
    - `summary_stats`: evaluation in text and figures as produced by scripts/evaluate_traces.py
    - `trace_csvs`: individual traces in plain text files, with predicted and ground truth label columns
    - `trace_plots`: visualized traces with two colored strips denoting predicted (upper) and ground
    truth (lower) classifications.
    from the FRETboard report by opening it in a browser and clicking "Download csv" under transition rate matrix.
   - `experiment_2_fb`: FRETboard results experiment 2
     - ...
   
## Rerunning analysis

To re-run the data evaluation, download this repo and cd to the root folder.
First install the included conda environment:
```
conda install -f scripts/env.yml
```
The analysis can then be rerun (from the root folder) from the command line:
```
bash execute_analysis.sh
```
Of course, the FRETboard results can be replaced by your own if you would like
to redo the analysis from scratch. In that case, remember to adhere to the 
structure of the 'eval' folder as described above.