Mercurial > repos > george-weingart > lefse
comparison run_lefse.py @ 0:e7cd19afda2e draft
Lefse
| author | george-weingart |
|---|---|
| date | Tue, 13 May 2014 21:57:00 -0400 |
| parents | |
| children |
comparison
equal
deleted
inserted
replaced
| -1:000000000000 | 0:e7cd19afda2e |
|---|---|
| 1 #!/usr/bin/env python | |
| 2 | |
| 3 import os,sys,math,pickle | |
| 4 from lefse import * | |
| 5 | |
| 6 def read_params(args): | |
| 7 parser = argparse.ArgumentParser(description='LEfSe 1.0') | |
| 8 parser.add_argument('input_file', metavar='INPUT_FILE', type=str, help="the input file") | |
| 9 parser.add_argument('output_file', metavar='OUTPUT_FILE', type=str, | |
| 10 help="the output file containing the data for the visualization module") | |
| 11 parser.add_argument('-o',dest="out_text_file", metavar='str', type=str, default="", | |
| 12 help="set the file for exporting the result (only concise textual form)") | |
| 13 parser.add_argument('-a',dest="anova_alpha", metavar='float', type=float, default=0.05, | |
| 14 help="set the alpha value for the Anova test (default 0.05)") | |
| 15 parser.add_argument('-w',dest="wilcoxon_alpha", metavar='float', type=float, default=0.05, | |
| 16 help="set the alpha value for the Wilcoxon test (default 0.05)") | |
| 17 parser.add_argument('-l',dest="lda_abs_th", metavar='float', type=float, default=2.0, | |
| 18 help="set the threshold on the absolute value of the logarithmic LDA score (default 2.0)") | |
| 19 parser.add_argument('--nlogs',dest="nlogs", metavar='int', type=int, default=3, | |
| 20 help="max log ingluence of LDA coeff") | |
| 21 parser.add_argument('--verbose',dest="verbose", metavar='int', choices=[0,1], type=int, default=0, | |
| 22 help="verbose execution (default 0)") | |
| 23 parser.add_argument('--wilc',dest="wilc", metavar='int', choices=[0,1], type=int, default=1, | |
| 24 help="wheter to perform the Wicoxon step (default 1)") | |
| 25 parser.add_argument('-r',dest="rank_tec", metavar='str', choices=['lda','svm'], type=str, default='lda', | |
| 26 help="select LDA or SVM for effect size (default LDA)") | |
| 27 parser.add_argument('--svm_norm',dest="svm_norm", metavar='int', choices=[0,1], type=int, default=1, | |
| 28 help="whether to normalize the data in [0,1] for SVM feature waiting (default 1 strongly suggested)") | |
| 29 parser.add_argument('-b',dest="n_boots", metavar='int', type=int, default=30, | |
| 30 help="set the number of bootstrap iteration for LDA (default 30)") | |
| 31 parser.add_argument('-e',dest="only_same_subcl", metavar='int', type=int, default=0, | |
| 32 help="set whether perform the wilcoxon test only among the subclasses with the same name (default 0)") | |
| 33 parser.add_argument('-c',dest="curv", metavar='int', type=int, default=0, | |
| 34 help="set whether perform the wilcoxon test ing the Curtis's approach [BETA VERSION] (default 0)") | |
| 35 parser.add_argument('-f',dest="f_boots", metavar='float', type=float, default=0.67, | |
| 36 help="set the subsampling fraction value for each bootstrap iteration (default 0.66666)") | |
| 37 parser.add_argument('-s',dest="strict", choices=[0,1,2], type=int, default=0, | |
| 38 help="set the multiple testing correction options. 0 no correction (more strict, default), 1 correction for independent comparisons, 2 correction for independent comparison") | |
| 39 # parser.add_argument('-m',dest="m_boots", type=int, default=5, | |
| 40 # help="minimum cardinality of classes in each bootstrapping iteration (default 5)") | |
| 41 parser.add_argument('--min_c',dest="min_c", metavar='int', type=int, default=10, | |
| 42 help="minimum number of samples per subclass for performing wilcoxon test (default 10)") | |
| 43 parser.add_argument('-t',dest="title", metavar='str', type=str, default="", | |
| 44 help="set the title of the analysis (default input file without extension)") | |
| 45 parser.add_argument('-y',dest="multiclass_strat", choices=[0,1], type=int, default=0, | |
| 46 help="(for multiclass tasks) set whether the test is performed in a one-against-one ( 1 - more strict!) or in a one-against-all setting ( 0 - less strict) (default 0)") | |
| 47 args = parser.parse_args() | |
| 48 | |
| 49 params = vars(args) | |
| 50 if params['title'] == "": params['title'] = params['input_file'].split("/")[-1].split('.')[0] | |
| 51 return params | |
| 52 | |
| 53 | |
| 54 | |
| 55 if __name__ == '__main__': | |
| 56 init() | |
| 57 params = read_params(sys.argv) | |
| 58 feats,cls,class_sl,subclass_sl,class_hierarchy = load_data(params['input_file']) | |
| 59 kord,cls_means = get_class_means(class_sl,feats) | |
| 60 wilcoxon_res = {} | |
| 61 kw_n_ok = 0 | |
| 62 nf = 0 | |
| 63 for feat_name,feat_values in feats.items(): | |
| 64 if params['verbose']: | |
| 65 print "Testing feature",str(nf),": ",feat_name, | |
| 66 nf += 1 | |
| 67 kw_ok,pv = test_kw_r(cls,feat_values,params['anova_alpha'],sorted(cls.keys())) | |
| 68 if not kw_ok: | |
| 69 if params['verbose']: print "\tkw ko" | |
| 70 del feats[feat_name] | |
| 71 wilcoxon_res[feat_name] = "-" | |
| 72 continue | |
| 73 if params['verbose']: print "\tkw ok\t", | |
| 74 | |
| 75 if not params['wilc']: continue | |
| 76 kw_n_ok += 1 | |
| 77 res_wilcoxon_rep = test_rep_wilcoxon_r(subclass_sl,class_hierarchy,feat_values,params['wilcoxon_alpha'],params['multiclass_strat'],params['strict'],feat_name,params['min_c'],params['only_same_subcl'],params['curv']) | |
| 78 wilcoxon_res[feat_name] = str(pv) if res_wilcoxon_rep else "-" | |
| 79 if not res_wilcoxon_rep: | |
| 80 if params['verbose']: print "wilc ko" | |
| 81 del feats[feat_name] | |
| 82 elif params['verbose']: print "wilc ok\t" | |
| 83 | |
| 84 if len(feats) > 0: | |
| 85 print "Number of significantly discriminative features:", len(feats), "(", kw_n_ok, ") before internal wilcoxon" | |
| 86 if params['lda_abs_th'] < 0.0: | |
| 87 lda_res,lda_res_th = dict([(k,0.0) for k,v in feats.items()]), dict([(k,v) for k,v in feats.items()]) | |
| 88 else: | |
| 89 if params['rank_tec'] == 'lda': lda_res,lda_res_th = test_lda_r(cls,feats,class_sl,params['n_boots'],params['f_boots'],params['lda_abs_th'],0.0000000001,params['nlogs']) | |
| 90 elif params['rank_tec'] == 'svm': lda_res,lda_res_th = test_svm(cls,feats,class_sl,params['n_boots'],params['f_boots'],params['lda_abs_th'],0.0,params['svm_norm']) | |
| 91 else: lda_res,lda_res_th = dict([(k,0.0) for k,v in feats.items()]), dict([(k,v) for k,v in feats.items()]) | |
| 92 else: | |
| 93 print "Number of significantly discriminative features:", len(feats), "(", kw_n_ok, ") before internal wilcoxon" | |
| 94 print "No features with significant differences between the two classes" | |
| 95 lda_res,lda_res_th = {},{} | |
| 96 outres = {} | |
| 97 outres['lda_res_th'] = lda_res_th | |
| 98 outres['lda_res'] = lda_res | |
| 99 outres['cls_means'] = cls_means | |
| 100 outres['cls_means_kord'] = kord | |
| 101 outres['wilcox_res'] = wilcoxon_res | |
| 102 print "Number of discriminative features with abs LDA score >",params['lda_abs_th'],":",len(lda_res_th) | |
| 103 save_res(outres,params["output_file"]) |