Mercurial > repos > george-weingart > lefse
comparison home/ubuntu/lefse_to_export/lefse.py @ 1:db64b6287cd6 draft
Modified datatypes
author | george-weingart |
---|---|
date | Wed, 20 Aug 2014 16:56:51 -0400 |
parents | |
children |
comparison
equal
deleted
inserted
replaced
0:e7cd19afda2e | 1:db64b6287cd6 |
---|---|
1 import os,sys,math,pickle | |
2 import random as lrand | |
3 import rpy2.robjects as robjects | |
4 import argparse | |
5 import numpy | |
6 #import svmutil | |
7 | |
8 def init(): | |
9 lrand.seed(1982) | |
10 robjects.r('library(splines)') | |
11 robjects.r('library(stats4)') | |
12 robjects.r('library(survival)') | |
13 robjects.r('library(mvtnorm)') | |
14 robjects.r('library(modeltools)') | |
15 robjects.r('library(coin)') | |
16 robjects.r('library(MASS)') | |
17 | |
18 def get_class_means(class_sl,feats): | |
19 means = {} | |
20 clk = class_sl.keys() | |
21 for fk,f in feats.items(): | |
22 means[fk] = [numpy.mean((f[class_sl[k][0]:class_sl[k][1]])) for k in clk] | |
23 return clk,means | |
24 | |
25 def save_res(res,filename): | |
26 with open(filename, 'w') as out: | |
27 for k,v in res['cls_means'].items(): | |
28 out.write(k+"\t"+str(math.log(max(max(v),1.0),10.0))+"\t") | |
29 if k in res['lda_res_th']: | |
30 for i,vv in enumerate(v): | |
31 if vv == max(v): | |
32 out.write(str(res['cls_means_kord'][i])+"\t") | |
33 break | |
34 out.write(str(res['lda_res'][k])) | |
35 else: out.write("\t") | |
36 out.write( "\t" + res['wilcox_res'][k]+"\n") | |
37 | |
38 def load_data(input_file, nnorm = False): | |
39 with open(input_file, 'rb') as inputf: | |
40 inp = pickle.load(inputf) | |
41 if nnorm: return inp['feats'],inp['cls'],inp['class_sl'],inp['subclass_sl'],inp['class_hierarchy'],inp['norm'] | |
42 else: return inp['feats'],inp['cls'],inp['class_sl'],inp['subclass_sl'],inp['class_hierarchy'] | |
43 | |
44 def load_res(input_file): | |
45 with open(input_file, 'rb') as inputf: | |
46 inp = pickle.load(inputf) | |
47 return inp['res'],inp['params'],inp['class_sl'],inp['subclass_sl'] | |
48 | |
49 | |
50 def test_kw_r(cls,feats,p,factors): | |
51 robjects.globalenv["y"] = robjects.FloatVector(feats) | |
52 for i,f in enumerate(factors): | |
53 robjects.globalenv['x'+str(i+1)] = robjects.FactorVector(robjects.StrVector(cls[f])) | |
54 fo = "y~x1" | |
55 #for i,f in enumerate(factors[1:]): | |
56 # if f == "subclass" and len(set(cls[f])) <= len(set(cls["class"])): continue | |
57 # if len(set(cls[f])) == len(cls[f]): continue | |
58 # fo += "+x"+str(i+2) | |
59 kw_res = robjects.r('kruskal.test('+fo+',)$p.value') | |
60 return float(tuple(kw_res)[0]) < p, float(tuple(kw_res)[0]) | |
61 | |
62 def test_rep_wilcoxon_r(sl,cl_hie,feats,th,multiclass_strat,mul_cor,fn,min_c,comp_only_same_subcl,curv=False): | |
63 comp_all_sub = not comp_only_same_subcl | |
64 tot_ok = 0 | |
65 alpha_mtc = th | |
66 all_diff = [] | |
67 for pair in [(x,y) for x in cl_hie.keys() for y in cl_hie.keys() if x < y]: | |
68 dir_cmp = "not_set" # | |
69 l_subcl1, l_subcl2 = (len(cl_hie[pair[0]]), len(cl_hie[pair[1]])) | |
70 if mul_cor != 0: alpha_mtc = th*l_subcl1*l_subcl2 if mul_cor == 2 else 1.0-math.pow(1.0-th,l_subcl1*l_subcl2) | |
71 ok = 0 | |
72 curv_sign = 0 | |
73 first = True | |
74 for i,k1 in enumerate(cl_hie[pair[0]]): | |
75 br = False | |
76 for j,k2 in enumerate(cl_hie[pair[1]]): | |
77 if not comp_all_sub and k1[len(pair[0]):] != k2[len(pair[1]):]: | |
78 ok += 1 | |
79 continue | |
80 cl1 = feats[sl[k1][0]:sl[k1][1]] | |
81 cl2 = feats[sl[k2][0]:sl[k2][1]] | |
82 med_comp = False | |
83 if len(cl1) < min_c or len(cl2) < min_c: | |
84 med_comp = True | |
85 sx,sy = numpy.median(cl1),numpy.median(cl2) | |
86 if cl1[0] == cl2[0] and len(set(cl1)) == 1 and len(set(cl2)) == 1: | |
87 tres, first = False, False | |
88 elif not med_comp: | |
89 robjects.globalenv["x"] = robjects.FloatVector(cl1+cl2) | |
90 robjects.globalenv["y"] = robjects.FactorVector(robjects.StrVector(["a" for a in cl1]+["b" for b in cl2])) | |
91 pv = float(robjects.r('pvalue(wilcox_test(x~y,data=data.frame(x,y)))')[0]) | |
92 tres = pv < alpha_mtc*2.0 | |
93 if first: | |
94 first = False | |
95 if not curv and ( med_comp or tres ): | |
96 dir_cmp = sx < sy | |
97 #if sx == sy: br = True | |
98 elif curv: | |
99 dir_cmp = None | |
100 if med_comp or tres: | |
101 curv_sign += 1 | |
102 dir_cmp = sx < sy | |
103 else: br = True | |
104 elif not curv and med_comp: | |
105 if ((sx < sy) != dir_cmp or sx == sy): br = True | |
106 elif curv: | |
107 if tres and dir_cmp == None: | |
108 curv_sign += 1 | |
109 dir_cmp = sx < sy | |
110 if tres and dir_cmp != (sx < sy): | |
111 br = True | |
112 curv_sign = -1 | |
113 elif not tres or (sx < sy) != dir_cmp or sx == sy: br = True | |
114 if br: break | |
115 ok += 1 | |
116 if br: break | |
117 if curv: diff = curv_sign > 0 | |
118 else: diff = (ok == len(cl_hie[pair[1]])*len(cl_hie[pair[0]])) # or (not comp_all_sub and dir_cmp != "not_set") | |
119 if diff: tot_ok += 1 | |
120 if not diff and multiclass_strat: return False | |
121 if diff and not multiclass_strat: all_diff.append(pair) | |
122 if not multiclass_strat: | |
123 tot_k = len(cl_hie.keys()) | |
124 for k in cl_hie.keys(): | |
125 nk = 0 | |
126 for a in all_diff: | |
127 if k in a: nk += 1 | |
128 if nk == tot_k-1: return True | |
129 return False | |
130 return True | |
131 | |
132 | |
133 | |
134 def contast_within_classes_or_few_per_class(feats,inds,min_cl,ncl): | |
135 ff = zip(*[v for n,v in feats.items() if n != 'class']) | |
136 cols = [ff[i] for i in inds] | |
137 cls = [feats['class'][i] for i in inds] | |
138 if len(set(cls)) < ncl: | |
139 return True | |
140 for c in set(cls): | |
141 if cls.count(c) < min_cl: | |
142 return True | |
143 cols_cl = [x for i,x in enumerate(cols) if cls[i] == c] | |
144 for i,col in enumerate(zip(*cols_cl)): | |
145 if (len(set(col)) <= min_cl and min_cl > 1) or (min_cl == 1 and len(set(col)) <= 1): | |
146 return True | |
147 return False | |
148 | |
149 def test_lda_r(cls,feats,cl_sl,boots,fract_sample,lda_th,tol_min,nlogs): | |
150 fk = feats.keys() | |
151 means = dict([(k,[]) for k in feats.keys()]) | |
152 feats['class'] = list(cls['class']) | |
153 clss = list(set(feats['class'])) | |
154 for uu,k in enumerate(fk): | |
155 if k == 'class': continue | |
156 ff = [(feats['class'][i],v) for i,v in enumerate(feats[k])] | |
157 for c in clss: | |
158 if len(set([float(v[1]) for v in ff if v[0] == c])) > max(float(feats['class'].count(c))*0.5,4): continue | |
159 for i,v in enumerate(feats[k]): | |
160 if feats['class'][i] == c: | |
161 feats[k][i] = math.fabs(feats[k][i] + lrand.normalvariate(0.0,max(feats[k][i]*0.05,0.01))) | |
162 rdict = {} | |
163 for a,b in feats.items(): | |
164 if a == 'class' or a == 'subclass' or a == 'subject': | |
165 rdict[a] = robjects.StrVector(b) | |
166 else: rdict[a] = robjects.FloatVector(b) | |
167 robjects.globalenv["d"] = robjects.DataFrame(rdict) | |
168 lfk = len(feats[fk[0]]) | |
169 rfk = int(float(len(feats[fk[0]]))*fract_sample) | |
170 f = "class ~ "+fk[0] | |
171 for k in fk[1:]: f += " + " + k.strip() | |
172 ncl = len(set(cls['class'])) | |
173 min_cl = int(float(min([cls['class'].count(c) for c in set(cls['class'])]))*fract_sample*fract_sample*0.5) | |
174 min_cl = max(min_cl,1) | |
175 pairs = [(a,b) for a in set(cls['class']) for b in set(cls['class']) if a > b] | |
176 | |
177 for k in fk: | |
178 for i in range(boots): | |
179 means[k].append([]) | |
180 for i in range(boots): | |
181 for rtmp in range(1000): | |
182 rand_s = [lrand.randint(0,lfk-1) for v in range(rfk)] | |
183 if not contast_within_classes_or_few_per_class(feats,rand_s,min_cl,ncl): break | |
184 rand_s = [r+1 for r in rand_s] | |
185 means[k][i] = [] | |
186 for p in pairs: | |
187 robjects.globalenv["rand_s"] = robjects.IntVector(rand_s) | |
188 robjects.globalenv["sub_d"] = robjects.r('d[rand_s,]') | |
189 z = robjects.r('z <- suppressWarnings(lda(as.formula('+f+'),data=sub_d,tol='+str(tol_min)+'))') | |
190 robjects.r('w <- z$scaling[,1]') | |
191 robjects.r('w.unit <- w/sqrt(sum(w^2))') | |
192 robjects.r('ss <- sub_d[,-match("class",colnames(sub_d))]') | |
193 if 'subclass' in feats: | |
194 robjects.r('ss <- ss[,-match("subclass",colnames(ss))]') | |
195 if 'subject' in feats: | |
196 robjects.r('ss <- ss[,-match("subject",colnames(ss))]') | |
197 robjects.r('xy.matrix <- as.matrix(ss)') | |
198 robjects.r('LD <- xy.matrix%*%w.unit') | |
199 robjects.r('effect.size <- abs(mean(LD[sub_d[,"class"]=="'+p[0]+'"]) - mean(LD[sub_d[,"class"]=="'+p[1]+'"]))') | |
200 scal = robjects.r('wfinal <- w.unit * effect.size') | |
201 rres = robjects.r('mm <- z$means') | |
202 rowns = list(rres.rownames) | |
203 lenc = len(list(rres.colnames)) | |
204 coeff = [abs(float(v)) if not math.isnan(float(v)) else 0.0 for v in scal] | |
205 res = dict([(pp,[float(ff) for ff in rres.rx(pp,True)] if pp in rowns else [0.0]*lenc ) for pp in [p[0],p[1]]]) | |
206 for j,k in enumerate(fk): | |
207 gm = abs(res[p[0]][j] - res[p[1]][j]) | |
208 means[k][i].append((gm+coeff[j])*0.5) | |
209 res = {} | |
210 for k in fk: | |
211 m = max([numpy.mean([means[k][kk][p] for kk in range(boots)]) for p in range(len(pairs))]) | |
212 res[k] = math.copysign(1.0,m)*math.log(1.0+math.fabs(m),10) | |
213 return res,dict([(k,x) for k,x in res.items() if math.fabs(x) > lda_th]) | |
214 | |
215 | |
216 def test_svm(cls,feats,cl_sl,boots,fract_sample,lda_th,tol_min,nsvm): | |
217 return NULL | |
218 """ | |
219 fk = feats.keys() | |
220 clss = list(set(cls['class'])) | |
221 y = [clss.index(c)*2-1 for c in list(cls['class'])] | |
222 xx = [feats[f] for f in fk] | |
223 if nsvm: | |
224 maxs = [max(v) for v in xx] | |
225 mins = [min(v) for v in xx] | |
226 x = [ dict([(i+1,(v-mins[i])/(maxs[i]-mins[i])) for i,v in enumerate(f)]) for f in zip(*xx)] | |
227 else: x = [ dict([(i+1,v) for i,v in enumerate(f)]) for f in zip(*xx)] | |
228 | |
229 lfk = len(feats[fk[0]]) | |
230 rfk = int(float(len(feats[fk[0]]))*fract_sample) | |
231 mm = [] | |
232 | |
233 best_c = svmutil.svm_ms(y, x, [pow(2.0,i) for i in range(-5,10)],'-t 0 -q') | |
234 for i in range(boots): | |
235 rand_s = [lrand.randint(0,lfk-1) for v in range(rfk)] | |
236 r = svmutil.svm_w([y[yi] for yi in rand_s], [x[xi] for xi in rand_s], best_c,'-t 0 -q') | |
237 mm.append(r[:len(fk)]) | |
238 m = [numpy.mean(v) for v in zip(*mm)] | |
239 res = dict([(v,m[i]) for i,v in enumerate(fk)]) | |
240 return res,dict([(k,x) for k,x in res.items() if math.fabs(x) > lda_th]) | |
241 """ |