-
Notifications
You must be signed in to change notification settings - Fork 0
/
Thomas-Agilent.Rmd
461 lines (380 loc) · 21.2 KB
/
Thomas-Agilent.Rmd
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
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
---
title: "Microplastic fingerprinting"
output:
md_document:
variant: markdown_github
---
```{r setup, include=FALSE}
knitr::opts_knit$set(root.dir = "C:/Users/huyng/OneDrive - Toronto Metropolitan University/Microplastic/Microplastic-Fingerprinting")
knitr::opts_chunk$set(echo = FALSE)
knitr::opts_chunk$set(out.width='750px', dpi=200)
```
## Documentation
This repo is accompanying the publication: "Computational fingerprinting workflow for environmental source tracking of microplastic based on representative additives"
Demo of each ML algorithms is shown below.
## Data processing
You can include R code in the document as follows:
```{r class.source = 'fold-hide', echo = FALSE, message = FALSE, warning = FALSE}
# Loading Packages --------------------------------------------------------
library(ggplot2)
library(vegan)
library(readxl)
library(tidyverse)
library(dplyr)
library(data.table)
library(writexl)
library(tidyr)
library(grid)
library(gridExtra)
library(plotly)
library(stats)
library(FactoMineR)
library(factoextra)
library(compositions)
library(ggforce)
library(latticeExtra)
library(cluster)
library(cowplot)
library(RColorBrewer)
library(reshape2)
library(corrplot)
'%notin%' <- Negate('%in%')
```
```{r class.source = 'fold-hide', echo = FALSE, message = FALSE, warning = FALSE}
# STEP 1.1: Data import --------------------------------------------
polymer <- readxl::read_xlsx(paste0(getwd(), "/TMU_Polymer_trainingsetA_Thomas_crookes.xlsx"),
.name_repair = "minimal" # No name repair or checks, beyond basic existence,
)
group_columns <- function(df) {
# Get the unique column names
unique_names <- unique(names(df))
# Initialize an empty list to store grouped column names
grouped_columns <- list()
# Loop through unique column names
for (name in unique_names) {
# Find columns with the same name
matching_columns <- which(names(df) == name)
# If there are duplicates, group them
if (length(matching_columns) > 1) {
grouped_columns[[name]] <- matching_columns
}
}
return(grouped_columns)
}
# Create a list of columns with the same name (aka. replicates of the same measurement)
column_groups <- group_columns(polymer)
df_list <- list()
count <- 1
# For each element in the list, make a dataframe that contains wavenumber, sample_name and values
for (i in 1:length(column_groups)) {
temp <- polymer[, c(1, column_groups[i][[1]])]
for (j in 2:ncol(temp)) {
colnames(temp)[j] <- paste0(colnames(temp)[j], "_rep", j-1)
}
temp <- temp %>% pivot_longer(cols = 2:ncol(.), names_to = "sample", values_to = "values")
df_list[[count]] <- temp
count <- count + 1
}
# Combine them all into big df
grand_df <- do.call(rbind, df_list) %>%
mutate(polymer = ifelse(grepl("ULDPE", sample, ignore.case = TRUE), "ULDPE",
ifelse(grepl("LLDPE1", sample, ignore.case = TRUE), "LLDPE1",
ifelse(grepl("LLDPE2", sample, ignore.case = TRUE), "LLDPE2",
ifelse(grepl("LDPE1", sample,, ignore.case = TRUE), "LDPE1",
ifelse(grepl("LDPE2", sample, ignore.case = TRUE), "LDPE2",
ifelse(grepl("MDPE", sample, ignore.case = TRUE), "MDPE",
ifelse(grepl("HDPE1", sample, ignore.case = TRUE), "HDPE1",
ifelse(grepl("HDPE2", sample,ignore.case = TRUE), "HDPE2",
ifelse(grepl("PP", sample,ignore.case = TRUE), "PP",
ifelse(grepl("PEST", sample,ignore.case = TRUE), "PEST",
ifelse(grepl("PET1", sample,ignore.case = TRUE), "PET1",
ifelse(grepl("PET2", sample,ignore.case = TRUE), "PET2",
ifelse(grepl("EVA", sample,ignore.case = TRUE), "EVA",
ifelse(grepl("ABS", sample,ignore.case = TRUE), "ABS",
ifelse(grepl("EPS", sample,ignore.case = TRUE), "EPS",
ifelse(grepl("PS", sample,ignore.case = TRUE), "PS",
ifelse(grepl("PA6", sample,ignore.case = TRUE), "PA6",
ifelse(grepl("PA66", sample,ignore.case = TRUE), "PA66",
ifelse(grepl("PVC1", sample,ignore.case = TRUE), "PVC1",
ifelse(grepl("PVC2", sample, ignore.case = TRUE), "PVC2",
ifelse(grepl("CR", sample, ignore.case = TRUE), "CR", "CA")))))))))))))))))))))) %>%
mutate(day = ifelse(grepl("day 0", sample, ignore.case = TRUE), 0,
ifelse(grepl("day 14", sample, ignore.case = TRUE), 14,
ifelse(grepl("day 30", sample, ignore.case = TRUE), 30,
ifelse(grepl("day 7", sample, ignore.case = TRUE), 7,
ifelse(grepl("day 1", sample, ignore.case = TRUE), 1,
ifelse(grepl("day 21", sample, ignore.case = TRUE), 21, 3))))))) %>%
filter(polymer %notin% "CA") %>%
pivot_wider(names_from = `Wavenumber (cm{¹)`, values_from = values) %>%
mutate(polymer_day = paste0(polymer, "_", day)) %>%
relocate(polymer_day, .after = 1)
# writexl::write_xlsx(grand_df, path = paste0(getwd(), "/Python/Thomas_Polymer_grand_df.xlsx"))
```
# Remove day 1 from analysis because of technical error
```{r , echo=FALSE, warning = FALSE, message=FALSE}
grand_df <- grand_df %>%
filter(day != 1)
```
# Exploratory data analysis
### Differences in each polymer - legend (Day 0 to Day 30)
```{r , echo=FALSE, warning = FALSE, message=FALSE}
input <- grand_df %>% mutate(day = factor(day, levels = unique(grand_df$day)))
plotdat <- input %>%
pivot_longer(cols = 5:ncol(.), names_to = "Wavenumber", values_to = "Values") %>%
mutate(Wavenumber = as.numeric(Wavenumber)) %>%
group_by(polymer, day, Wavenumber) %>%
summarise(across(Values, base::mean))
plotlist <- list()
i <- 1
for (pol in unique(plotdat$polymer)) {
subdf <- plotdat %>%
filter(polymer == pol)
print(ggplot(subdf, aes(x = Wavenumber, y = Values, color = day)) +
geom_point(size = 0.5) +
labs(title = pol,
x = NULL,
y = NULL) +
theme_minimal(base_size = 18) +
theme(panel.grid.minor.x = element_blank(),
panel.grid.minor.y = element_blank(),
plot.title = element_text(hjust = 0.5)) +
scale_y_continuous(limits = c(0, round(max(plotdat$Values))), expand = c(0,0))+
guides(colour = guide_legend(override.aes = list(size = 10))))
i <- i + 1
}
# legend <- cowplot::get_legend(ggplot(subdf, aes(x = Wavenumber, y = Values, color = day)) +
# geom_point(size = 0.5) +
# labs(title = pol,
# x = NULL,
# y = NULL) +
# theme_minimal(base_size = 18) +
# theme(panel.grid.minor.x = element_blank(),
# panel.grid.minor.y = element_blank()) +
# guides(colour = guide_legend(override.aes = list(size = 10))))
#
# gridExtra::grid.arrange(grobs = plotlist, ncol = 3,
# bottom = textGrob("Wavenumber (cm{¹)",
# gp = gpar(fontsize = 20)),
# right = legend
# )
```
### Differences in each Day 0 to Day 30 across all polymer
```{r , echo=FALSE, warning = FALSE, message=FALSE}
input <- grand_df %>% mutate(polymer = factor(polymer, levels = unique(grand_df$polymer)))
# polymer_colors <- grDevices::colorRampPalette(brewer.pal(12, "Set2"))(21)
# polymer_color_map <- setNames(polymer_colors, unique(grand_df$polymer))
plotdat <- input %>%
pivot_longer(cols = 5:ncol(.), names_to = "Wavenumber", values_to = "Values") %>%
mutate(Wavenumber = as.numeric(Wavenumber)) %>%
group_by(polymer, day, Wavenumber) %>%
summarise(across(Values, base::mean))
#
# plotlist <- list()
# i <- 1
for (day in unique(plotdat$day)) {
subdf <- plotdat %>%
filter(day == 30)
# plotlist[[i]] <-
ggplot(subdf, aes(x = Wavenumber, y = Values, color = polymer)) +
geom_point(size = 0.5) +
# scale_color_manual(values = polymer_color_map) +
labs(title = paste0("Day ", 30),
x = NULL,
y = NULL) +
theme_minimal(base_size = 18) +
theme(legend.position = "bottom",
panel.grid.minor.x = element_blank(),
panel.grid.minor.y = element_blank(),
plot.title = element_text(hjust = 0.5)) +
guides(colour = guide_legend(override.aes = list(size = 10), ncol=5)) +
scale_y_continuous(limits = c(0, round(max(plotdat$Values))), expand = c(0,0))
# i <- i + 1
}
```
# Correlation analysis
```{r , echo=FALSE, warning = FALSE, message=FALSE}
data <- grand_df %>%
mutate(polymer = factor(polymer, levels = unique(grand_df$polymer))) %>%
mutate(day = factor(day, levels = unique(grand_df$day))) %>%
pivot_longer(cols = 5:ncol(.), names_to = "Wavenumber", values_to = "Values") %>%
mutate(Wavenumber = as.numeric(Wavenumber)) %>%
group_by(polymer, day, Wavenumber) %>%
summarise(across(Values, base::mean)) %>%
pivot_wider(names_from = Wavenumber, values_from = Values) %>%
mutate(polymer_day = paste0(polymer, "_Day_", day)) %>%
column_to_rownames(., var = "polymer_day") %>%
filter(polymer %in% c("ABS", "PVC1", "PVC2", "CR"))
# c("HDPE2", "HDPE1", "LDPE1", "LDPE2", "LLDPE1", "MDPE", "PP", "PA6"))
# c("PS", "PP", "HDPE1", "LDPE1", "LDPE2", "LLDPE1", "MDPE", "EPS", "HDPE2"))
cor_matrix <- cor(t(data[, 3:ncol(data)]))
# Extract the upper triangle of the correlation matrix
cor_matrix[lower.tri(cor_matrix)] <- NA
# Melt the correlation matrix, excluding NA values
melted_cor_matrix <- melt(cor_matrix, na.rm = TRUE)
# Create the heatmap using ggplot2
ggplot(data = melted_cor_matrix, aes(x = Var1, y = Var2, fill = value)) +
geom_tile(color = "white") +
scale_fill_gradient2(low = "blue", high = "red", mid = "white",
midpoint = 0, limit = c(-1, 1), space = "Lab",
name="Correlation") +
theme_minimal() +
theme(axis.text.x = element_text(angle = 90, vjust = 1,
size = 12, hjust = 1),
axis.text.y = element_text(size = 12)) +
coord_fixed() +
labs(title = "", x = "", y = "")
```
# Statistical fingerprinting
## ANOVA
(Update 29th May 2024:) aov() function ==> Error in model.frame.default(formula = as.formula(paste(paste(setdiff(names(input), :
variable lengths differ (found for 'polymer')
```{r , echo=FALSE, warning = FALSE, message=FALSE}
input <- subset(grand_df, select = -c(sample, day, polymer_day))
summary(aov(as.formula(paste(paste(setdiff(names(input), "polymer"), collapse = "+"), " ~ polymer")), data = input))
```
## PCA
(Update 3rd may 2024;) PCA of both zero and global minimum provides no distinctive clusters
```{r , echo=FALSE, warning = FALSE, message=FALSE}
# PCA
input <- grand_df %>%
filter(polymer %notin% c("EVA", "ABS", "CR", "LDPE1", "LLDPE2"))
res.pca <- FactoMineR::PCA(
input %>%
select(-c("sample", "day", "polymer", "polymer_day")),
scale.unit = FALSE,
graph = FALSE)
# Scree plot
# fviz_screeplot(res.pca, ncp=10)
# Biplot
factoextra::fviz_pca_biplot(res.pca,
geom = c("point", "text"),
label = "none",
invisible = "var",
repel = TRUE,
labelsize = 10,
habillage = factor(input$polymer),
addEllipses = TRUE,
ellipse.level=0.95,
ggtheme = ggplot2::theme_minimal(base_size = 20),
title = "",
# xlim = c(-0.5, 0.5),
# ylim = c(-0.25, 0.2)
) +
guides(fill=guide_legend(ncol=5)) +
# if has error "Too few points to calculate an ellipse"
# ggforce::geom_mark_ellipse(aes(fill = Groups,
# color = Groups),
# label.buffer = unit(40, 'mm')) +
theme(legend.position = 'bottom',
panel.grid.minor.x = element_blank(),
panel.grid.minor.y = element_blank()
)
```
## HCA
If we filter unique compounds by File (compounds appear in at least 2 files), then all File in df in gc_hplc_list are the same. That's why here we used gc_hplc_list[[1]]
This exercise does not seems to affect the HCa. Keep in mind that for all 3 cases of data combinations: none of them has compounds that occur in all samples. NONE of THEM!!!!!
### Clustering of samples
(Update 3rd May 2024:) With zero and global minimum imputation, I saw that the two mixture was clustered within 2 out of 3 of its components (Woodbridge foam and new tire rubber)
```{r echo=FALSE, warning=FALSE, message=FALSE}
input <- grand_df # %>%
# filter(day == 0)
hc_df <- input %>%
column_to_rownames(., var = "sample") %>%
select(-c("day", "polymer", "polymer_day"))
```
### Use cut_height to see broader grouping of samples
```{r, echo=FALSE, warning = FALSE, message=FALSE}
## Dissimilarity Indices Calculated by vegan::vegdist()
hca_samp <- stats::hclust(vegan::vegdist(hc_df,
method = "canberra")) # Since our data is continous -> canberra // manhattan // aitchison // robust.aitchison
plot(hca_samp,
labels = input$polymer,
hang = -1,
main = "")
rect.hclust(hca_samp, h = round(hca_samp$height[828], digits = 4), border = "red")
hca_dat <- data.frame(group = stats::cutree(hca_samp, h = round(hca_samp$height[828], digits = 4))) %>%
arrange(group) %>%
rownames_to_column(., var = "sample") %>%
mutate(polymer = ifelse(grepl("ULDPE", sample, ignore.case = TRUE), "ULDPE",
ifelse(grepl("LLDPE1", sample, ignore.case = TRUE), "LLDPE1",
ifelse(grepl("LLDPE2", sample, ignore.case = TRUE), "LLDPE2",
ifelse(grepl("LDPE1", sample,, ignore.case = TRUE), "LDPE1",
ifelse(grepl("LDPE2", sample, ignore.case = TRUE), "LDPE2",
ifelse(grepl("MDPE", sample, ignore.case = TRUE), "MDPE",
ifelse(grepl("HDPE1", sample, ignore.case = TRUE), "HDPE1",
ifelse(grepl("HDPE2", sample,ignore.case = TRUE), "HDPE2",
ifelse(grepl("PP", sample,ignore.case = TRUE), "PP",
ifelse(grepl("PEST", sample,ignore.case = TRUE), "PEST",
ifelse(grepl("PET1", sample,ignore.case = TRUE), "PET1",
ifelse(grepl("PET2", sample,ignore.case = TRUE), "PET2",
ifelse(grepl("EVA", sample,ignore.case = TRUE), "EVA",
ifelse(grepl("ABS", sample,ignore.case = TRUE), "ABS",
ifelse(grepl("EPS", sample,ignore.case = TRUE), "EPS",
ifelse(grepl("PS", sample,ignore.case = TRUE), "PS",
ifelse(grepl("PA6", sample,ignore.case = TRUE), "PA6",
ifelse(grepl("PA66", sample,ignore.case = TRUE), "PA66",
ifelse(grepl("PVC1", sample,ignore.case = TRUE), "PVC1",
ifelse(grepl("PVC2", sample, ignore.case = TRUE), "PVC2",
ifelse(grepl("CR", sample, ignore.case = TRUE), "CR", "CA")))))))))))))))))))))) %>%
mutate(day = ifelse(grepl("day 0", sample, ignore.case = TRUE), "day 0",
ifelse(grepl("day 14", sample, ignore.case = TRUE), "day 14",
ifelse(grepl("day 30", sample, ignore.case = TRUE), "day 30",
ifelse(grepl("day 7", sample, ignore.case = TRUE), "day 7",
ifelse(grepl("day 1", sample, ignore.case = TRUE), "day 1",
ifelse(grepl("day 21", sample, ignore.case = TRUE), "day 21", "day 3"))))))) %>%
filter(polymer %notin% "CA") %>%
mutate(polymer_day = paste0(polymer, "_", day)) %>%
mutate(polymer = factor(polymer, levels = unique(polymer))) %>%
mutate(polymer_day = factor(polymer_day, levels = unique(polymer_day))) %>%
mutate(group = factor(group, levels = unique(group))) %>%
mutate(day = factor(day, levels = unique(day)))
View(hca_dat)
# writexl::write_xlsx(hca_dat, path = paste0(getwd(), "/hca_dat.xlsx"))
# Plot the distribution of polymer occurrences for each HCA group
# ggplot(hca_dat, aes(x = group, fill = polymer)) +
# geom_bar(position = "dodge") +
# labs(title = "Distribution of Polymer Occurrences by Group", x = "Group", y = "Count") +
# scale_x_discrete(drop = FALSE) +
# theme_minimal()
# Plot the distribution of days for each HCA group
ggplot(hca_dat, aes(x = polymer_day)) +
geom_bar() +
facet_wrap(~ group, scales = "free_x") + # Create separate panels for each group
labs(title = "Distribution of Days by Group", x = "Day", y = "Count") +
theme_minimal(base_size = 15) +
theme(axis.text.x = element_text(angle = 90, hjust = 1))
```
## K-means Clustering
```{r, echo=FALSE, warning = FALSE, message=FALSE}
tot.withinss <- list()
for (n_clus in 1:nrow(df_percentage)) {
set.seed(round(runif(2:20, 0, 999)))
km.out <- stats::kmeans(hc_df, 20, nstart = 50,
algorithm = "Hartigan-Wong" # "Hartigan-Wong", "Lloyd", "MacQueen"
)
tot.withinss[[n_clus]] <- km.out$tot.withinss
print(paste0("The number of centers is ", 20, " and total within-cluster sum of square is ", km.out$tot.withinss))
}
plot(df, col = (stats::kmeans(df, 171, nstart = 50)$cluster + 1) ,
main = "K- Means Clustering Results with K = 2",
xlab = "", ylab = "", pch = 20, cex = 2)
```
## UMAP clustering
```{r, echo=FALSE, warning = FALSE, message=FALSE}
library(umap)
input <- grand_df #%>%
# filter(day == 30)
features <- subset(input, select = -c(sample, day, polymer, polymer_day))
umap <- umap(features, n_components = 3,
method = 'naive',
metric= "pearson2", # euclidean, manhattan, cosine, pearson, pearson2
alpha = 0.00001, gamma = 0.0001)
layout <- cbind(data.frame(umap[["layout"]]), input$day)
umap_plot <- plot_ly(layout, x = ~X1, y = ~X2, z = ~X3,
color = ~input$day) %>%
add_markers() %>%
layout(scene = list(xaxis = list(title = 'x-axis'),
yaxis = list(title = 'y-axis'),
zaxis = list(title = 'z-axis')))
umap_plot
```