Skip to content

Latest commit

 

History

History
74 lines (61 loc) · 3.47 KB

README.md

File metadata and controls

74 lines (61 loc) · 3.47 KB
Raw

Phenylketonuria landscape analysis v 1.9.0

Pipeline:

  1. Apply Gaussian model and extract features
  2. Perform QC (rmse - var)
  3. Detect non-responders
  4. Subpopulation detection

Run pipeline

Install packages

numpy
pandas
scipy
matplotlib
openpyxl
tqdm
scikit-learn      # needed for prediction only

To run the pipeline use the script gm.py, which requres experimant files as *.xlsm.

Example 1) run the pipeline for one experimant

python gm.py --input_dir Data/experiments_all/20240305_EXP13.xlsm

Example 2) run the pipeline for all experimants in a directory (e.g. Data/experiments_all)

python gm.py --input_dir Data/experiments_all

Promp terminal parameters:

  • --input_dir: input directiry. Either a *.xlsm file or a folder containing them.
  • --param_dir: directory to parameters json file (see below).
  • --no_plot: no plot will be created if apecified.

A parameter .json file is also needed to specify the parameters. By default the json file should be in the same folder as the gm.py named gm_parameters.json. You can change this by the parameter --param_dir.

Parameters in the json file:

  • save_image2d_dir: Directory to save 2D landscapes; set it as "" to prevent saving.
  • save_image3d_dir: Directory to save 3D plots; set it as "" to prevent saving.
  • save_feature_dir: Directory to save feature csv file; set it as "" to prevent saving.
  • tag: Name tag for the feature .csv file.
  • qc_thr_rmse: QC threshold for RMSE; default=[0.2, 0.25]
  • qc_thr_n_peaks: QC threshold for N Peacks; default=[5, 8]
  • qc_thr_variation: QC threshold for Variations; default=[0.1, 0.25]
  • elev: Elevation of the camera in the 3D plots; default=30,
  • azim: Angle of the camera in the 3D plots (in degrees);default=-120,
  • nbins: Number of bins used for the interpolation (mesh grid of size nbins x nbins);default=1000,
  • sm_method: Method used for the interpolation and smoothing ("regular_grid" or "linear_ndi");default='regular_grid',
  • rescale: Binary value to rescale the data with respect to the WT;default=True,
  • max_val_scale: Maximum value of the non-rescaled data;default=10000,
  • info_box: Binary value to add an information box to the plots;default=True,
  • max_val: Binary value to add the maximum value of the data to the information box;default=True,
  • peak_coords: Binary value to add the coordinates of the peaks to the information box;default=True,
  • fifty_coords: Binary value to add the coordinates of the 50% of the maximum value to the information ;default=True,
  • plot_replicates: Binary value to plot and save data for all the replicates;default=False
  • plot_extra": Binary value to plot and save the data in the log scale';default=False

Predict for new data

When we have new data samples, the script predict.py can be used to predict the lables of clusters for each sample.

Example)

python predict.py --train_data_dir clustering/data/clustering_result.csv --new_data_dir gm_output/features/extracted_features.csv

Promp terminal parameters:

  • --train_data_dir: Train data directory; the output of the clustering pipeline provided in R.
  • --new_data_dir: New data directory; the output of the gm.py for the train data.
  • --k: Number of neighbours k in KNN.