-
Notifications
You must be signed in to change notification settings - Fork 0
/
S01-ondisk.R
182 lines (147 loc) · 6.14 KB
/
S01-ondisk.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
library("MSnbase")
prep_data <- function(n,
f = msdata::proteomics(full.names = TRUE, pattern = "20141210"),
data_dir = "./data",
clean = TRUE) {
if (clean && dir.exists(data_dir))
unlink(data_dir, recursive = TRUE)
if (!dir.exists(data_dir))
dir.create(data_dir)
fls <- paste0(data_dir, "/f", seq_len(n), ".mzML")
stopifnot(all(file.link(f, fls)))
fls
}
fls <- prep_data(100)
## ----------------------------------------------------------------------
## BENCHMARK:
## - Reading time (in seconds) for a 1, 5, 10, ... files (default
## settings, i.e. 6103 MS2 spectra for in memory and 1431 MS1 + 6103
## MS2 spectra for on disk)
## - Memory consumption, compareing on disk and in memory for 1, 5,
## 10, ... files
## ----------------------------------------------------------------------
n <- c(1, 5, 10)
time_sz <- lapply(n, function(i) {
f <- fls[seq_len(i)]
time <- c(in_mem = system.time(x_mem <- readMSData(f, mode = "inMemory"))[["elapsed"]],
on_disk = system.time(x_dsk <- readMSData(f, mode = "onDisk"))[["elapsed"]])
sz <- c(in_mem = pryr::object_size(x_mem),
on_disk = pryr::object_size(x_dsk))
cbind(time, sz, n = i)
})
save(time_sz, file = "bench_time_sz.rda")
## Repeat reading time in triplicates
time_2 <- data.frame(
n = rep(n, each = 3),
time_mem = rep(NA, 9),
time_dsk = rep(NA, 9))
for (i in seq_len(nrow(time_2))) {
n <- time_2$n[i]
f <- fls[seq_len(n)]
time_2[i, "time_mem"] <- system.time(x_mem <- readMSData(f, mode = "inMemory"))[["elapsed"]]
time_2[i, "time_dsk"] <- system.time(x_dsk <- readMSData(f, mode = "onDisk"))[["elapsed"]]
print(time_2)
}
save(time_2, file = "bench_time_2.rda")
## time_2 %>%
## mutate(ratio = time_mem/time_dsk) %>%
## pull(ratio) %>%
## mean
## 5.367308
## ----------------------------------------------------------------------
## BENCHMARK: filtering
## ----------------------------------------------------------------------
library(tidyverse)
library(microbenchmark)
## Documenting the filters used:
x <- readMSData(fls[1], mode = "onDisk")
chr <- chromatogram(x)
## 1. keep 1000 < rtimes < 3000
plot(chr, col = "black")
## 2. 50% of MS2 spectra with most intense precuror intensity
median(precursorIntensity(x), na.rm = TRUE)
## 3. Focus on TMT
plot(x_dsk[[1227]], full = TRUE, reporters = TMT6)
TMT6
range(mz(TMT6))
x_dsk <- readMSData(fls[1], mode = "onDisk", msLevel = 2L)
x_mem <- readMSData(fls[1], mode = "inMemory", msLevel = 2L)
filter_ms <- function(x, i = NULL) {
x <- x %>%
filterRt(c(1000, 3000)) %>%
filterMz(120, 135)
x <- x[precursorIntensity(x) > 11e6, ]
if (!is.null(i))
x <- x[seq_len(i), ]
x
}
t_filt <- microbenchmark(dsk = filter_ms(x_dsk),
mem = filter_ms(x_mem),
times = 10)
save(t_filt, file = "bench_t_filt.rda")
autoplot(t_filt)
## ----------------------------------------------------------------------
## BENCHMARK: accessing 1, 10, ..., all spectra
## ----------------------------------------------------------------------
t_access <- microbenchmark(
access_mem_1 = spectra(x_mem[1]),
access_mem_10 = spectra(x_mem[1:10]),
access_mem_100 = spectra(x_mem[1:100]),
access_mem_1000 = spectra(x_mem[1:1000]),
access_mem_5000 = spectra(x_mem[1:5000]),
access_mem_all = spectra(x_mem),
access_dsk_1 = spectra(x_dsk[1]),
access_dsk_10 = spectra(x_dsk[1:10]),
access_dsk_100 = spectra(x_dsk[1:100]),
access_dsk_1000 = spectra(x_dsk[1:1000]),
access_dsk_5000 = spectra(x_dsk[1:5000]),
access_dsk_all = spectra(x_dsk),
times = 10)
save(t_access, file = "bench_t_access.rda")
tibble(time = microbenchmark:::convert_to_unit(t_access$time, "ms"),
expr = as.character(t_access$expr)) %>%
mutate(n = sub("^access_.+_", "", expr)) %>%
mutate(n = sub("all", length(x_dsk), n)) %>%
mutate(mode = if_else(grepl("mem", expr), "inMem", "onDisk")) %>%
ggplot(aes(x = n, y = time, fill = log10(as.numeric(n)))) +
ggplot2::geom_violin() +
ggplot2::scale_y_log10() +
ylab("Time [millisec]") +
xlab("Number of spectra (out of 6103)") +
facet_wrap(~ mode) +
theme(legend.position = "none")
## ----------------------------------------------------------------------
## BENCHMARK: All together for 1, 5000, all spectra
## ----------------------------------------------------------------------
t_full <- microbenchmark(mem_all = readMSData(fls[1], mode = "inMemory", msLevel = 2L) %>%
filter_ms() %>%
spectra(),
mem_1 = readMSData(fls[1], mode = "inMemory", msLevel = 2L) %>%
filter_ms(i = 1) %>%
spectra(),
mem_1000 = readMSData(fls[1], mode = "inMemory", msLevel = 2L) %>%
filter_ms(i = 1000) %>%
spectra(),
dsk_all = readMSData(fls[1], mode = "onDisk", msLevel = 2L) %>%
filter_ms() %>%
spectra(),
dsk_1 = readMSData(fls[1], mode = "onDisk", msLevel = 2L) %>%
filter_ms(i = 1) %>%
spectra(),
dsk_1000 = readMSData(fls[1], mode = "onDisk", msLevel = 2L) %>%
filter_ms(i = 1000) %>%
spectra(),
times = 10)
save(t_full, file = "bench_t_full.rda")
tibble(time = microbenchmark:::convert_to_unit(t_full$time, "s"),
expr = as.character(t_full$expr)) %>%
mutate(n = sub("^.+_", "", expr)) %>%
mutate(n = sub("all", length(x_dsk), n)) %>%
mutate(mode = if_else(grepl("mem", expr), "inMem", "onDisk")) %>%
ggplot(aes(x = n, y = time, fill = log10(as.numeric(n)))) +
ggplot2::geom_violin() +
ggplot2::scale_y_log10() +
ylab("Time [seconds]") +
xlab("Number of spectra (out of 6103)") +
facet_wrap(~ mode) +
theme(legend.position = "none")