-
Notifications
You must be signed in to change notification settings - Fork 0
/
explore-reduce-method-comp.R
237 lines (148 loc) · 6.38 KB
/
explore-reduce-method-comp.R
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
library(data.table)
setDTthreads(threads = 0)
## Created in merge-compsfixNulls-wide_g.R
load("./plots-glms/wide_g_method_comp.RData")
df <- wide_g_method_comp
colnames(df)
## This is dangerous
colsnum <- colnames(df)[25:52]
df[, (colsnum) := lapply(.SD, as.numeric), .SDcols = colsnum]
## All of them generate
## 1: In lapply(.SD, as.numeric) : NAs introduced by coercion
## 1460 NAs
sum(is.na(df$"CBN -vs- MHN"))
100 * sum(is.na(df$"CBN -vs- MHN"))/nrow(df) ## 0.06
lapply(df[, 25:52], function(x) sum(is.na(x)))
somenas <- which(is.na(df$"CBN -vs- MHN"))[c(1, 15, 1000, 1460)]
df[somenas, 1:27]
## OK, only those from the 16 cases where MCCBN had an NA
dfnona <- df[!is.na(df$min_js), ]
lapply(dfnona[, 25:52], function(x) sum(is.na(x)))
## 0.77
cor.test(dfnona$"CBN -vs- MHN", dfnona$"MCCBN -vs- MHN", method = "spearman")
## -0.0024
cor.test(dfnona$"CBN -vs- MCCBN", dfnona$"MCCBN -vs- MHN", method = "spearman")
## -0.1853
cor.test(dfnona$"CBN -vs- MHN", dfnona$"CBN -vs- MCCBN", method = "spearman")
## 0.863
cor.test(dfnona$"CAPRESE -vs- MHN", dfnona$"MHN -vs- OT", method = "spearman")
## 0.8237
cor.test(dfnona$"CAPRESE -vs- CBN", dfnona$"CBN -vs- OT", method = "spearman")
## - 0.25
cor.test(dfnona$"CAPRESE -vs- MHN", dfnona$"CAPRESE -vs- OT", method = "spearman")
## 0.45
cor.test(dfnona$"MHN -vs- MHN_td", dfnona$"CBN -vs- CBN_td", method = "spearman")
## 0.6211
cor.test(dfnona$"MCCBN -vs- MCCBN_td", dfnona$"CBN -vs- CBN_td", method = "spearman")
## 0.35
cor.test(dfnona$"MHN -vs- MHN_td", dfnona$"MCCBN -vs- MCCBN_td", method = "spearman")
## 0.3353
cor.test(dfnona$"CBN_td -vs- MHN", dfnona$"CBN -vs- MHN_td", method = "spearman")
## 0.19
cor.test(dfnona$"MHN -vs- MHN_td", dfnona$"CBN -vs- MHN", method = "spearman")
## Need to rename columns
colnames(dfnona) <- gsub(" -vs- ", "..vs..", colnames(dfnona))
## TD_comp_CE: time-discretized compared to CompetingExponentials
## Based on the figures, MCCBN_td is not as good a representative of the
## "_td" cases. Its average performance, when MHN_th and CBN_td do a great
## job, is generally poorer.
dfnona[, `:=`(CBN_MHN = (1/3) * (CBN..vs..MHN + CBN..vs..MCCBN + MCCBN..vs..MHN),
OT_CAPRESE = CAPRESE..vs..OT,
## TD = (1/3) * (CBN_td..vs..MHN_td + CBN_td..vs..MCCBN_td + MCCBN_td..vs..MHN_td),
TD = CBN_td..vs..MHN_td,
## TD_comp_CE = (1/3) * (CBN..vs..CBN_td + MCCBN..vs..MCCBN_td + MHN..vs..MHN_td),
TD_comp_CE = (1/2) * (CBN..vs..CBN_td + MHN..vs..MHN_td),
CBN_comp_OT = (1/6) * (CBN..vs..OT + MCCBN..vs..OT + MHN..vs..OT +
CAPRESE..vs..CBN + CAPRESE..vs..MCCBN + CAPRESE..vs..MHN),
CE = CBN..vs..MHN,
MHN_CE_TD = MHN..vs..MHN_td,
CBN_CE_TD = CBN..vs..CBN_td
) ]
nrow(dfnona)
summary(dfnona)
## ## remove the 16 missing. Nope, don't
## dfnona_clean <- dfnona[!is.na(CBN_MHN), ]
dfnona_clean <- dfnona
nrow(dfnona_clean)
summary(dfnona_clean)
## Keep only variables we will need
dfnona_clean[, `:=`( typeLandscape = typeLandscape_f)]
levels(dfnona_clean$typeLandscape_f) <- c("Rep", "LM", "RMF")
with(dfnona_clean, table(typeLandscape_f, typeLandscape))
dfnona_clean <- dfnona_clean[, .(id, sourceGenotype, replicate,
size_split, detect,
min_js, propLocalMax,
sourceGenotype_nMut, numGenes,
sampledProp,
CE,
MHN_CE_TD,
CBN_CE_TD,
CBN_MHN, OT_CAPRESE, TD,
TD_comp_CE, CBN_comp_OT,
typeLandscape_f
)]
method_comp_fit <- dfnona_clean
save(file = "method_comp_fit.RData", method_comp_fit,
compress = FALSE)
## All of them being similar is captured by, well, all of them being small
## 0.06
cor.test(method_comp_fit$CBN_MHN, method_comp_fit$OT_CAPRESE, method = "spearman")
## 0.7
cor.test(method_comp_fit$CBN_MHN, method_comp_fit$TD, method = "spearman")
## -0.05
cor.test(method_comp_fit$OT_CAPRESE, method_comp_fit$TD, method = "spearman")
## -0.15
cor.test(method_comp_fit$TD_comp_CE, method_comp_fit$TD, method = "spearman")
## 0.08
cor.test(method_comp_fit$TD_comp_CE, method_comp_fit$CBN_MHN, method = "spearman")
## 0.44
cor.test(method_comp_fit$TD_comp_CE, method_comp_fit$OT_CAPRESE, method = "spearman")
## 0.17
cor.test(method_comp_fit$TD_comp_CE, method_comp_fit$CBN_comp_OT, method = "spearman")
## 0.73
cor.test(method_comp_fit$CBN_MHN, method_comp_fit$CBN_comp_OT, method = "spearman")
## 0.26
cor.test(method_comp_fit$OT_CAPRESE, method_comp_fit$CBN_comp_OT, method = "spearman")
## In case we wanted all vars, which we dont
## dfnona_all_vars <- dfnona
## dfnona_all_vars <- dfnona_all_vars[!is.na(TD), ]
## colsToDelete <- colnames(dfnona_all_vars)[c(8:10, 12:20, 22:23)]
## dfnona_all_vars[, (colsToDelete) := NULL]
## method_comp_rf_fit <- dfnona_all_vars
## save(file = "method_comp_rf_fit.RData", method_comp_rf_fit,
## compress = FALSE)
## ###############
## ##### Older stuff, pasting other info
## load("data_for_weighted_glmertree_plots.RData")
## nrow(data_for_weighted_glmertree_plots)
## nrow(df)
## df2 <- df
## max_g_7 <- 6
## max_g_10 <- 9
## df2 <- df2[
## ((numGenes == 7) & (sourceGenotype_nMut <= max_g_7))
## |
## ((numGenes == 10) & (sourceGenotype_nMut <= max_g_10)),
## ]
## nrow(df2)
## any(is.na(df2$min_js))
## df2 <- df2[!is.na(df2$min_js), ]
## nrow(df2)
## ## Make sure same names
## setnames(data_for_weighted_glmertree_plots,
## old = c("sample_size"),
## new = c("size_split"))
## samecols <- intersect(colnames(df2), colnames(data_for_weighted_glmertree_plots))
## df3 <- merge.data.table(df2, data_for_weighted_glmertree_plots,
## by = samecols,
## all.x = TRUE, all.y = TRUE)
## dim(df3)
## dim(df2)
## sum(is.na(df3$"CBN -vs- MHN"))
## sum(is.na(df3$"CBN -vs- MHN"))/nrow(df3) ## 34%
## somenas <- which(is.na(df3$"CBN -vs- MHN"))[c(1, 15, 1000, 200000, 815000, 815398)]
## df3[somenas, ]
## ## with(df, cor.test("CBN -vs- MCCBN", "CBN -vs- MHN"))
## df3somenas <- df3[somenas, ]
## save(file = "df3_somenas.RData", df3somenas)
## summary(df3[, 25:52])