Coverage for biobb_ml/clustering/dbscan.py: 85%

1#!/usr/bin/env python3 

2 

3"""Module containing the DBSCANClustering class and the command line interface.""" 

4import argparse 

5import pandas as pd 

6from biobb_common.generic.biobb_object import BiobbObject 

7from sklearn.preprocessing import StandardScaler 

8from sklearn.cluster import DBSCAN 

9from biobb_common.configuration import settings 

10from biobb_common.tools import file_utils as fu 

11from biobb_common.tools.file_utils import launchlogger 

12from biobb_ml.clustering.common import check_input_path, check_output_path, getHeader, getIndependentVars, getIndependentVarsList, hopkins, plotCluster 

13 

14 

15class DBSCANClustering(BiobbObject): 

16 """ 

17 | biobb_ml DBSCANClustering 

18 | Wrapper of the scikit-learn DBSCAN method. 

19 | Clusters a given dataset. Visit the `DBSCAN documentation page <https://scikit-learn.org/stable/modules/generated/sklearn.cluster.DBSCAN.html>`_ in the sklearn official website for further information. 

20 

21 Args: 

22 input_dataset_path (str): Path to the input dataset. File type: input. `Sample file <https://github.com/bioexcel/biobb_ml/raw/master/biobb_ml/test/data/clustering/dataset_dbscan.csv>`_. Accepted formats: csv (edam:format_3752). 

23 output_results_path (str): Path to the clustered dataset. File type: output. `Sample file <https://github.com/bioexcel/biobb_ml/raw/master/biobb_ml/test/reference/clustering/ref_output_results_dbscan.csv>`_. Accepted formats: csv (edam:format_3752). 

24 output_plot_path (str) (Optional): Path to the clustering plot. File type: output. `Sample file <https://github.com/bioexcel/biobb_ml/raw/master/biobb_ml/test/reference/clustering/ref_output_plot_dbscan.png>`_. Accepted formats: png (edam:format_3603). 

25 properties (dic - Python dictionary object containing the tool parameters, not input/output files): 

26 * **predictors** (*dict*) - ({}) Features or columns from your dataset you want to use for fitting. You can specify either a list of columns names from your input dataset, a list of columns indexes or a range of columns indexes. Formats: { "columns": ["column1", "column2"] } or { "indexes": [0, 2, 3, 10, 11, 17] } or { "range": [[0, 20], [50, 102]] }. In case of mulitple formats, the first one will be picked. 

27 * **eps** (*float*) - (0.5) [0~10|0.1] The maximum distance between two samples for one to be considered as in the neighborhood of the other. 

28 * **min_samples** (*int*) - (5) [1~100|1] The number of samples (or total weight) in a neighborhood for a point to be considered as a core point. This includes the point itself. 

29 * **metric** (*str*) - ("euclidean") The metric to use when calculating distance between instances in a feature array. Values: cityblock (Compute the City Block -Manhattan- distance), cosine (Compute the Cosine distance between 1-D arrays), euclidean (Computes the Euclidean distance between two 1-D arrays), l1, l2, manhattan (Compute the Manhattan distance), braycurtis (Compute the Bray-Curtis distance between two 1-D arrays), canberra (Compute the Canberra distance between two 1-D arrays), chebyshev (Compute the Chebyshev distance), correlation (Compute the correlation distance between two 1-D arrays), dice (Compute the Dice dissimilarity between two boolean 1-D arrays), hamming (Compute the Hamming distance between two 1-D arrays), jaccard (Compute the Jaccard-Needham dissimilarity between two boolean 1-D arrays), kulsinski (Compute the Kulsinski dissimilarity between two boolean 1-D arrays), mahalanobis (Compute the Mahalanobis distance between two 1-D arrays), minkowski (Compute the Minkowski distance between two 1-D arrays), rogerstanimoto (Compute the Rogers-Tanimoto dissimilarity between two boolean 1-D arrays), russellrao (Compute the Russell-Rao dissimilarity between two boolean 1-D arrays), seuclidean (Return the standardized Euclidean distance between two 1-D arrays), sokalmichener (Compute the Sokal-Michener dissimilarity between two boolean 1-D arrays), sokalsneath (Compute the Sokal-Sneath dissimilarity between two boolean 1-D arrays), sqeuclidean (Compute the squared Euclidean distance between two 1-D arrays), yule (Compute the Yule dissimilarity between two boolean 1-D arrays). 

30 * **plots** (*list*) - (None) List of dictionaries with all plots you want to generate. Only 2D or 3D plots accepted. Format: [ { 'title': 'Plot 1', 'features': ['feat1', 'feat2'] } ]. 

31 * **scale** (*bool*) - (False) Whether or not to scale the input dataset. 

32 * **remove_tmp** (*bool*) - (True) [WF property] Remove temporal files. 

33 * **restart** (*bool*) - (False) [WF property] Do not execute if output files exist. 

34 

35 Examples: 

36 This is a use example of how to use the building block from Python:: 

37 

38 from biobb_ml.clustering.dbscan import dbscan 

39 prop = { 

40 'predictors': { 

41 'columns': [ 'column1', 'column2', 'column3' ] 

42 }, 

43 'eps': 1.4, 

44 'min_samples': 3, 

45 'metric': 'euclidean', 

46 'plots': [ 

47 { 

48 'title': 'Plot 1', 

49 'features': ['feat1', 'feat2'] 

50 } 

51 ] 

52 } 

53 dbscan(input_dataset_path='/path/to/myDataset.csv', 

54 output_results_path='/path/to/newTable.csv', 

55 output_plot_path='/path/to/newPlot.png', 

56 properties=prop) 

57 

58 Info: 

59 * wrapped_software: 

60 * name: scikit-learn DBSCAN 

61 * version: >=0.24.2 

62 * license: BSD 3-Clause 

63 * ontology: 

64 * name: EDAM 

65 * schema: http://edamontology.org/EDAM.owl 

66 

67 """ 

68 

69 def __init__(self, input_dataset_path, output_results_path, 

70 output_plot_path=None, properties=None, **kwargs) -> None: 

71 properties = properties or {} 

72 

73 # Call parent class constructor 

74 super().__init__(properties) 

75 self.locals_var_dict = locals().copy() 

76 

77 # Input/Output files 

78 self.io_dict = { 

79 "in": {"input_dataset_path": input_dataset_path}, 

80 "out": {"output_results_path": output_results_path, "output_plot_path": output_plot_path} 

81 } 

82 

83 # Properties specific for BB 

84 self.predictors = properties.get('predictors', {}) 

85 self.eps = properties.get('eps', .5) 

86 self.min_samples = properties.get('min_samples', 5) 

87 self.metric = properties.get('metric', 'euclidean') 

88 self.plots = properties.get('plots', []) 

89 self.scale = properties.get('scale', False) 

90 self.properties = properties 

91 

92 # Check the properties 

93 self.check_properties(properties) 

94 self.check_arguments() 

95 

96 def check_data_params(self, out_log, err_log): 

97 """ Checks all the input/output paths and parameters """ 

98 self.io_dict["in"]["input_dataset_path"] = check_input_path(self.io_dict["in"]["input_dataset_path"], "input_dataset_path", out_log, self.__class__.__name__) 

99 self.io_dict["out"]["output_results_path"] = check_output_path(self.io_dict["out"]["output_results_path"], "output_results_path", False, out_log, self.__class__.__name__) 

100 if self.io_dict["out"]["output_plot_path"]: 

101 self.io_dict["out"]["output_plot_path"] = check_output_path(self.io_dict["out"]["output_plot_path"], "output_plot_path", True, out_log, self.__class__.__name__) 

102 

103 @launchlogger 

104 def launch(self) -> int: 

105 """Execute the :class:`DBSCANClustering <clustering.dbscan.DBSCANClustering>` clustering.dbscan.DBSCANClustering object.""" 

106 

107 # check input/output paths and parameters 

108 self.check_data_params(self.out_log, self.err_log) 

109 

110 # Setup Biobb 

111 if self.check_restart(): 

112 return 0 

113 self.stage_files() 

114 

115 # load dataset 

116 fu.log('Getting dataset from %s' % self.io_dict["in"]["input_dataset_path"], self.out_log, self.global_log) 

117 if 'columns' in self.predictors: 

118 labels = getHeader(self.io_dict["in"]["input_dataset_path"]) 

119 skiprows = 1 

120 else: 

121 labels = None 

122 skiprows = None 

123 data = pd.read_csv(self.io_dict["in"]["input_dataset_path"], header=None, sep="\\s+|;|:|,|\t", engine="python", skiprows=skiprows, names=labels) 

124 

125 # the features are the predictors 

126 predictors = getIndependentVars(self.predictors, data, self.out_log, self.__class__.__name__) 

127 fu.log('Predictors: [%s]' % (getIndependentVarsList(self.predictors)), self.out_log, self.global_log) 

128 

129 # Hopkins test 

130 H = hopkins(predictors) 

131 fu.log('Performing Hopkins test over dataset. H = %f' % H, self.out_log, self.global_log) 

132 

133 # scale dataset 

134 if self.scale: 

135 fu.log('Scaling dataset', self.out_log, self.global_log) 

136 scaler = StandardScaler() 

137 predictors = scaler.fit_transform(predictors) 

138 

139 # create a DBSCAN object with self.clusters clusters 

140 model = DBSCAN(eps=self.eps, min_samples=self.min_samples, metric=self.metric) 

141 # fit the data 

142 model.fit(predictors) 

143 

144 # create a copy of data, so we can see the clusters next to the original data 

145 clusters = data.copy() 

146 # predict the cluster for each observation 

147 clusters['cluster'] = model.fit_predict(predictors) 

148 

149 fu.log('Calculating results\n\nCLUSTERING TABLE\n\n%s\n' % clusters, self.out_log, self.global_log) 

150 

151 # get number of clusters discarding outliers 

152 clstrs = set(clusters['cluster']) 

153 if -1 in clstrs: 

154 clstrs.remove(-1) 

155 fu.log('Total of clusters computed by DBSCAN = %d' % len(clstrs), self.out_log, self.global_log) 

156 

157 outliers = clusters['cluster'].tolist().count(-1) 

158 op = (outliers / len(clusters['cluster'].tolist())) * 100 

159 fu.log('Total of outliers = %d (%.2f%%)' % (outliers, op), self.out_log, self.global_log) 

160 

161 # save results 

162 fu.log('Saving results to %s' % self.io_dict["out"]["output_results_path"], self.out_log, self.global_log) 

163 clusters.to_csv(self.io_dict["out"]["output_results_path"], index=False, header=True, float_format='%.3f') 

164 

165 if self.io_dict["out"]["output_plot_path"] and self.plots: 

166 new_plots = [] 

167 i = 0 

168 for plot in self.plots: 

169 if len(plot['features']) == 2 or len(plot['features']) == 3: 

170 new_plots.append(plot) 

171 i += 1 

172 if i == 6: 

173 break 

174 

175 plot = plotCluster(new_plots, clusters) 

176 fu.log('Saving output plot to %s' % self.io_dict["out"]["output_plot_path"], self.out_log, self.global_log) 

177 plot.savefig(self.io_dict["out"]["output_plot_path"], dpi=150) 

178 

179 # Copy files to host 

180 self.copy_to_host() 

181 

182 self.tmp_files.extend([ 

183 self.stage_io_dict.get("unique_dir") 

184 ]) 

185 self.remove_tmp_files() 

186 

187 self.check_arguments(output_files_created=True, raise_exception=False) 

188 

189 return 0 

190 

191 

192def dbscan(input_dataset_path: str, output_results_path: str, output_plot_path: str = None, properties: dict = None, **kwargs) -> int: 

193 """Execute the :class:`DBSCANClustering <clustering.dbscan.DBSCANClustering>` class and 

194 execute the :meth:`launch() <clustering.dbscan.DBSCANClustering.launch>` method.""" 

195 

196 return DBSCANClustering(input_dataset_path=input_dataset_path, 

197 output_results_path=output_results_path, 

198 output_plot_path=output_plot_path, 

199 properties=properties, **kwargs).launch() 

200 

201 

202def main(): 

203 """Command line execution of this building block. Please check the command line documentation.""" 

204 parser = argparse.ArgumentParser(description="Wrapper of the scikit-learn DBSCAN method.", formatter_class=lambda prog: argparse.RawTextHelpFormatter(prog, width=99999)) 

205 parser.add_argument('--config', required=False, help='Configuration file') 

206 

207 # Specific args of each building block 

208 required_args = parser.add_argument_group('required arguments') 

209 required_args.add_argument('--input_dataset_path', required=True, help='Path to the input dataset. Accepted formats: csv.') 

210 required_args.add_argument('--output_results_path', required=True, help='Path to the clustered dataset. Accepted formats: csv.') 

211 parser.add_argument('--output_plot_path', required=False, help='Path to the clustering plot. Accepted formats: png.') 

212 

213 args = parser.parse_args() 

214 args.config = args.config or "{}" 

215 properties = settings.ConfReader(config=args.config).get_prop_dic() 

216 

217 # Specific call of each building block 

218 dbscan(input_dataset_path=args.input_dataset_path, 

219 output_results_path=args.output_results_path, 

220 output_plot_path=args.output_plot_path, 

221 properties=properties) 

222 

223 

224if __name__ == '__main__': 

225 main()