# Norm-ISWSVR

## Introduction

- Norm-ISWSVR   can be used for large-scale untargeted & targeted metabolomics data   normalization，removing   the systematic errors. Such as batch effects，matrix effect， injection   volume，instrument   drift and et al.

- The main processing   is shown in the figure below:

> Besides,   this package supports `cross validation` and `hyper parameters tuning`.


Note:

1. Boolean   parameters `use_*` are configurable in using this package.

2. With color   label of the rectangular is the default process, is also the recommended   processing in our paper.

3. Patent has   been applied, strictly prohibited for commercial use

## Install

```

git clone   https://github.com/Dingxian666/NormISWSVR.git

cd Norm-ISWSVR

python setup.py   install

```

## Data   Preparation

You should   prepare your data as the following format:

- data.csv

> Examples:   [demo-has-MatchedIS.csv](./examples/MatchedIS/demo-has-MatchedIS.csv) or   [demo-no-MatchedIS.csv](./examples/NoMatchedIS/data-no-matchedIS.csv)

Note: The   `matchedIS` column is optional.

- sample.csv

> Examples:   [sample-info.csv](./examples/MatchedIS/sample-info.csv)

Note: The   `injection_order` & `batch` columns are optional.

## Usage   Examples

- default

```

import ISWSVR

norm_iswsvr =   ISWSVR.NormISWSVR(data_path='./examples/MatchedIS/demo-has-MatchedIS.csv',

                                 sample_path='./examples/MatchedIS/sample-info.csv',

                                 save_dir='./examples/results',

                                 has_matched_IS=True)

norm_iswsvr.iswsvr()

```

- cross   validation

```

import ISWSVR

norm_iswsvr =   ISWSVR.NormISWSVR(data_path='./examples/MatchedIS/demo-has-MatchedIS.csv',

                                 sample_path='./examples/MatchedIS/sample-info.csv',

                                 save_dir='./examples/results',

                                 has_matched_IS=True)

norm_iswsvr.run_with_cv_folds(fold_nums=5)

```

-   hyper-parameters tuning with grid searching

```

import ISWSVR

norm_iswsvr =   ISWSVR.NormISWSVR(data_path='./examples/MatchedIS/demo-has-MatchedIS.csv',

                                 sample_path='./examples/MatchedIS/sample-info.csv',

                                 save_dir='./examples/results',

                                 has_matched_IS=True)

norm_iswsvr.run_with_grid_search(fold_nums=5,

                                  result_save_path='./examples/grid_search_results.csv',

                                  svr_top_corr_k=[1,2,3,4,5],

                                  svr_gamma=['auto', 0.1, 0.2, 0.4, 0.5],

                                  svr_C=[0.1, 0.2, 0.5, 1.0],

                                  tuned_times=50)

```

**For more   functions, please refer the function help during using, thank you!**

Please cite our   paper if you find our work useful for your research: