-
Notifications
You must be signed in to change notification settings - Fork 2
/
05_Lindstrom_QC.Rmd
612 lines (473 loc) · 16.7 KB
/
05_Lindstrom_QC.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
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
---
title: "Lindstrom Quality Control"
output:
workflowr::wflow_html:
code_folding: hide
---
```{r knitr, include = FALSE}
DOCNAME = "05_Lindstrom_QC"
knitr::opts_chunk$set(autodep = TRUE,
cache = TRUE,
cache.path = paste0("cache/", DOCNAME, "/"),
cache.comments = FALSE,
echo = TRUE,
error = FALSE,
fig.align = "center",
fig.width = 10,
fig.height = 8,
message = FALSE,
warning = FALSE)
```
```{r libaries, cache = FALSE}
# scRNA-seq
library("scater")
# Matrices
library("Matrix")
# Plotting
library("cowplot")
# Presentation
library("glue")
library("knitr")
# Parallel
library("BiocParallel")
# Paths
library("here")
# Output
library("jsonlite")
# Tidyverse
library("tidyverse")
```
```{r source, cache = FALSE}
source(here("R/load.R"))
source(here("R/output.R"))
```
Introduction
============
In this document we are going to read in the Lindstrom human fetal kidney data,
produce various quality control plots and remove any low-quality cells or
uninformative genes.
```{r load}
genes.map <- read_tsv(here("data/genes.tsv"),
col_types = cols(
gene_id = col_character(),
gene_name = col_character()
)) %>%
as.data.frame()
sce <- loadTSVSCE(here("data/Lindstrom/GSM2741551_count-table-human16w.tsv"),
genes.map,
dataset = "Lindstrom",
org = "human",
add.anno = TRUE,
calc.qc = TRUE,
calc.cpm = TRUE,
pct.mt = TRUE,
pct.ribo = TRUE,
cell.cycle = TRUE,
sparse = TRUE,
bpparam = MulticoreParam(workers = 10),
verbose = TRUE)
colData(sce)$Sample <- "1"
sce <- normalise(sce)
sce <- runPCA(sce)
sce <- runTSNE(sce)
write_rds(sce, here("data/processed/Lindstrom_SCE_complete.Rds"))
```
QC plots
========
By cell
-------
### Total counts
Violin plots of the library size (total counts) for each of the samples.
```{r total-counts}
plotColData(sce, x = "Sample", y = "total_counts", colour_by = "Sample")
```
### Number of features by library size
Relationship between the total counts for each cell and the number of expressed
genes. We expect the number of genes to increase with the number of counts,
hopefully reaching saturation.
```{r count-features}
plotColData(sce, x = "total_counts", y = "total_features", colour_by = "Sample")
```
### PCA
PCA plots coloured by different variables.
```{r pca}
p1 <- plotPCA(sce, colour_by = "Sample", shape_by = "Sample") +
ggtitle("Sample")
p2 <- plotPCA(sce, colour_by = "log10_total_counts", shape_by = "Sample") +
ggtitle("Total Counts")
p3 <- plotPCA(sce, colour_by = "CellCycle", shape_by = "Sample") +
ggtitle("Cell Cycle")
p4 <- plotPCA(sce, colour_by = "pct_dropout", shape_by = "Sample") +
ggtitle("Dropout")
p5 <- plotPCA(sce, colour_by = "PctCountsMT", shape_by = "Sample") +
ggtitle("Mitochondrial Genes")
p6 <- plotPCA(sce, colour_by = "PctCountsRibo", shape_by = "Sample") +
ggtitle("Ribosomal Genes")
plot_grid(p1, p2, p3, p4, p5, p6, ncol = 2)
```
### t-SNE
t-SNE plots coloured by different variables.
```{r t-SNE}
p1 <- plotTSNE(sce, colour_by = "Sample", shape_by = "Sample") +
ggtitle("Sample")
p2 <- plotTSNE(sce, colour_by = "log10_total_counts", shape_by = "Sample") +
ggtitle("Total Counts")
p3 <- plotTSNE(sce, colour_by = "CellCycle", shape_by = "Sample") +
ggtitle("Cell Cycle")
p4 <- plotTSNE(sce, colour_by = "pct_dropout", shape_by = "Sample") +
ggtitle("Dropout")
p5 <- plotTSNE(sce, colour_by = "PctCountsMT", shape_by = "Sample") +
ggtitle("Mitochondrial Genes")
p6 <- plotTSNE(sce, colour_by = "PctCountsRibo", shape_by = "Sample") +
ggtitle("Ribosomal Genes")
plot_grid(p1, p2, p3, p4, p5, p6, ncol = 2)
```
### Explantory variables
Plots of the variance explained by various variables.
```{r exp-vars}
exp.vars <- c("Sample", "CellCycle", "log10_total_counts",
"log10_total_features", "pct_dropout",
"pct_counts_top_200_features", "PctCountsMT", "PctCountsRibo")
all.zero <- rowSums(as.matrix(counts(sce))) == 0
plotExplanatoryVariables(sce[!all.zero, ], variables = exp.vars)
```
Correlation between explanatory variables.
```{r exp-vars-pairs}
plotExplanatoryVariables(sce[!all.zero, ], variables = exp.vars,
method = "pairs")
```
### RLE
Relative Log Expression (RLE) plots. Ideally all boxes should be aligned and
have the size size.
```{r rle}
plotRLE(sce[!all.zero], list(logcounts = "logcounts", counts = "counts"),
c(TRUE, FALSE), colour_by = "Sample")
```
### Mitochondrial genes
Looking at the effect of mitchondrial genes. We define mitochondrial genes as
genes on the MT chromosome or with "mitochondrial" in the description.
```{r pct-mt}
plotColData(sce, x = "Sample", y = "PctCountsMT", colour_by = "Sample")
```
### Ribosomal genes
Looking at the effect of ribosomal genes. We define ribosomal genes as
genes with "ribosom" in the description.
```{r pct-ribo}
plotColData(sce, x = "Sample", y = "PctCountsRibo", colour = "Sample")
```
### Housekeeping genes
Plots of housekeeping genes. We may want to use these for filtering as a
proxy for the health of the cell.
```{r hk-exprs}
actb.id <- filter(data.frame(rowData(sce)), feature_symbol == "ACTB")[1, 1]
gapdh.id <- filter(data.frame(rowData(sce)), feature_symbol == "GAPDH")[1, 1]
key <- c("ACTB", "GAPDH")
names(key) <- c(actb.id, gapdh.id)
plotExpression(sce, c(actb.id, gapdh.id), colour_by = "Sample") +
scale_x_discrete(labels = key)
```
By gene
-------
### High expression
```{r high-expression}
plotHighestExprs(sce)
```
### Expression frequency by mean
```{r mean-expression-freq}
plotExprsFreqVsMean(sce)
```
### Total counts by num cells expressed
```{r gene-expression}
plotRowData(sce, x = "n_cells_counts", y = "log10_total_counts")
```
Filtering
=========
```{r add-filter}
colData(sce)$Filtered <- FALSE
```
To begin with we have `r ncol(sce)` cells with `r nrow(sce)` features from the
ENSEMBL annotation.
Cells
-----
### Quantification
Let's consider how many reads are assigned to features. We can plot the total
number of counts in each cell against the number of genes that are expressed.
Cells that have been filtered are shown as triangles.
```{r quantification}
thresh.h <- 3500
plotColData(sce, x = "total_counts", y = "total_features",
colour_by = "Sample", shape_by = "Filtered") +
geom_hline(yintercept = thresh.h, colour = "red", size = 1.5,
linetype = "dashed") +
xlab("Total counts") +
ylab("Total features") +
ggtitle("Quantification metrics")
```
Cells that express many genes are potential multiplets (multiple cells
captured in a single droplet). We will remove
`r sum(colData(sce)$total_features > thresh.h)` cells with more than
`r thresh.h` genes expressed.
```{r quantification-filter}
colData(sce)$Filtered <- colData(sce)$Filtered |
colData(sce)$total_features > thresh.h
```
We now have `r sum(!colData(sce)$Filtered)` cells.
### Mitochondrial genes
Over-expression of mitochondrial genes can be an indication that a cell is
stressed or damaged in some way. Let's have a look at the percentage of counts
that are assigned to mitchondrial genes.
```{r mt}
thresh.h <- 8
plotColData(sce, x = "Sample", y = "PctCountsMT", colour_by = "Sample",
shape_by = "Filtered") +
geom_hline(yintercept = thresh.h, colour = "red", size = 1.5,
linetype = "dashed") +
xlab("Sample") +
ylab("% counts MT") +
ggtitle("Mitochondrial genes")
```
Some of the cells show high proportions of MT counts. We will remove
`r sum(colData(sce)$PctCountsMT > thresh.h)` cells with greater than
`r thresh.h`% MT counts.
```{r mt-filter}
colData(sce)$Filtered <- colData(sce)$Filtered |
colData(sce)$PctCountsMT > thresh.h
```
That leaves `r sum(!colData(sce)$Filtered)` cells.
### Ribsomal genes
We can do a similar thing for ribosomal gene expression.
```{r ribosomal}
thresh.h <- 35
plotColData(sce, x = "Sample", y = "PctCountsRibo", colour_by = "Sample",
shape_by = "Filtered") +
geom_hline(yintercept = thresh.h, colour = "red", size = 1.5,
linetype = "dashed") +
xlab("Sample") +
ylab("% counts ribosomal") +
ggtitle("Ribosomal genes")
```
Some of the cells show high proportions of ribosomal counts. We will remove
`r sum(colData(sce)$PctCountsRibo > thresh.h)` cells with greater than
`r thresh.h`% ribosomal counts.
```{r ribosomal-filter}
colData(sce)$Filtered <- colData(sce)$Filtered |
colData(sce)$PctCountsRibo > thresh.h
```
That leaves `r sum(!colData(sce)$Filtered)` cells.
### Housekeeping
Similarly we can look at the expression of the "housekeeping" genes GAPDH and
ACTB.
```{r housekeeping}
thresh.h <- 1
thresh.v <- 1.5
plotExpression(sce, gapdh.id, x = actb.id, colour_by = "Sample",
shape_by = "Filtered") +
geom_hline(yintercept = thresh.h, colour = "red", size = 1.5,
linetype = "dashed") +
geom_vline(xintercept = thresh.v, colour = "red", size = 1.5,
linetype = "dashed") +
xlab("ACTB") +
ylab("GAPDH") +
ggtitle("Housekeepking genes") +
theme(
strip.background = element_blank(),
strip.text.x = element_blank()
)
```
We will remove cells where ACTB is expressed below `r thresh.v` or GAPDH is
expressed below `r thresh.h`. This removes
`r sum(exprs(sce)[actb.id, ] < thresh.v | exprs(sce)[gapdh.id, ] < thresh.h)`
cells.
```{r housekeeping-filter}
colData(sce)$Filtered <- colData(sce)$Filtered |
exprs(sce)[actb.id, ] < thresh.v |
exprs(sce)[gapdh.id, ] < thresh.h
```
```{r filter-cells}
sce <- sce[, !colData(sce)$Filtered]
```
After filtering we are left with `r sum(!colData(sce)$Filtered)` cells.
### Dimensionality reduction (filtered cells)
Now that we are relatively confident we have a set of good quality cells, let's
see what they look like in reduced dimensions.
#### PCA
```{r pca-filtered-cells}
sce <- runPCA(sce)
p1 <- plotPCA(sce, colour_by = "Sample", shape_by = "Sample") +
ggtitle("Sample")
p2 <- plotPCA(sce, colour_by = "log10_total_counts", shape_by = "Sample") +
ggtitle("Total Counts")
p3 <- plotPCA(sce, colour_by = "CellCycle", shape_by = "Sample") +
ggtitle("Cell Cycle")
p4 <- plotPCA(sce, colour_by = "pct_dropout", shape_by = "Sample") +
ggtitle("Dropout")
p5 <- plotPCA(sce, colour_by = "PctCountsMT", shape_by = "Sample") +
ggtitle("Mitochondrial Genes")
p6 <- plotPCA(sce, colour_by = "PctCountsRibo", shape_by = "Sample") +
ggtitle("Ribosomal Genes")
plot_grid(p1, p2, p3, p4, p5, p6, ncol = 2)
```
#### t-SNE
```{r tSNE-filtered-cells}
sce <- runTSNE(sce)
p1 <- plotTSNE(sce, colour_by = "Sample", shape_by = "Sample") +
ggtitle("Sample")
p2 <- plotTSNE(sce, colour_by = "log10_total_counts", shape_by = "Sample") +
ggtitle("Total Counts")
p3 <- plotTSNE(sce, colour_by = "CellCycle", shape_by = "Sample") +
ggtitle("Cell Cycle")
p4 <- plotTSNE(sce, colour_by = "pct_dropout", shape_by = "Sample") +
ggtitle("Dropout")
p5 <- plotTSNE(sce, colour_by = "PctCountsMT", shape_by = "Sample") +
ggtitle("Mitochondrial Genes")
p6 <- plotTSNE(sce, colour_by = "PctCountsRibo", shape_by = "Sample") +
ggtitle("Ribosomal Genes")
plot_grid(p1, p2, p3, p4, p5, p6, ncol = 2)
```
Genes
-----
### Expression
Some of the features we would never expect to see expressed in an RNA-seq
experiment. Before doing anything else we will remove the features that have
less than two counts across all cells.
```{r expression}
keep <- rowSums(counts(sce)) > 1
sce <- sce[keep]
```
This removes `r sum(!keep)` genes and leaves us with `r nrow(sce)`.
We will also going remove genes that are expressed in less than two individual
cells.
```{r two-cells}
keep <- rowSums(counts(sce) != 0) > 1
sce <- sce[keep, ]
```
This removes `r sum(!keep)` genes and leaves us with `r nrow(sce)`.
### HGNC genes
We are also going to filter out any genes that don't have HGNC symbols. These
are mostly pseudogenes and are unlikely to be informative.
```{r hgnc-genes}
keep <- !(rowData(sce)$hgnc_symbol == "") &
!(is.na(rowData(sce)$hgnc_symbol))
sce <- sce[keep, ]
```
This removes `r sum(!keep)` genes and leaves us with `r nrow(sce)`.
```{r dedup-hgnc}
dups <- which(duplicated(rowData(sce)$feature_symbol))
```
There are `r length(dups)` gene(s) with duplicate HGNC symbol names. For these
genes we will use an alternative symbol name. Once we have done this we can
rename the features using feature symbols instead of ENSEMBL IDs which will make
interpreting results easier.
```{r rename-features}
rowData(sce)[dups, "feature_symbol"] <- rowData(sce)[dups, "symbol"]
rownames(sce) <- rowData(sce)$feature_symbol
```
### Dimensionality reduction (filtered genes)
Let's see what our final dataset looks like in reduced dimensions.
#### PCA
```{r pca-filtered-genes}
sce <- runPCA(sce)
p1 <- plotPCA(sce, colour_by = "Sample", shape_by = "Sample") +
ggtitle("Sample")
p2 <- plotPCA(sce, colour_by = "log10_total_counts", shape_by = "Sample") +
ggtitle("Total Counts")
p3 <- plotPCA(sce, colour_by = "CellCycle", shape_by = "Sample") +
ggtitle("Cell Cycle")
p4 <- plotPCA(sce, colour_by = "pct_dropout", shape_by = "Sample") +
ggtitle("Dropout")
p5 <- plotPCA(sce, colour_by = "PctCountsMT", shape_by = "Sample") +
ggtitle("Mitochondrial Genes")
p6 <- plotPCA(sce, colour_by = "PctCountsRibo", shape_by = "Sample") +
ggtitle("Ribosomal Genes")
plot_grid(p1, p2, p3, p4, p5, p6, ncol = 2)
```
#### t-SNE
```{r tSNE-filtered-genes}
sce <- runTSNE(sce)
p1 <- plotTSNE(sce, colour_by = "Sample", shape_by = "Sample") +
ggtitle("Sample")
p2 <- plotTSNE(sce, colour_by = "log10_total_counts", shape_by = "Sample") +
ggtitle("Total Counts")
p3 <- plotTSNE(sce, colour_by = "CellCycle", shape_by = "Sample") +
ggtitle("Cell Cycle")
p4 <- plotTSNE(sce, colour_by = "pct_dropout", shape_by = "Sample") +
ggtitle("Dropout")
p5 <- plotTSNE(sce, colour_by = "PctCountsMT", shape_by = "Sample") +
ggtitle("Mitochondrial Genes")
p6 <- plotTSNE(sce, colour_by = "PctCountsRibo", shape_by = "Sample") +
ggtitle("Ribosomal Genes")
plot_grid(p1, p2, p3, p4, p5, p6, ncol = 2)
```
We now have a high-quality dataset for our analysis with `r nrow(sce)` genes
and `r ncol(sce)` cells. A median of `r median(colSums(counts(sce) != 0))` genes
are expressed in each cell.
Summary
=======
Parameters
----------
This table describes parameters used and set in this document.
```{r parameters}
params <- toJSON(list(
list(
Parameter = "total_features",
Value = 3500,
Description = "Maximum threshold for total features expressed"
),
list(
Parameter = "mt_counts",
Value = 8,
Description = "Maximum threshold for percentage counts mitochondrial"
),
list(
Parameter = "ribo_counts",
Value = 35,
Description = "Maximum threshold for percentage counts ribosomal"
),
list(
Parameter = "ACTB_expr",
Value = 1.5,
Description = "Minimum threshold for ACTB expression"
),
list(
Parameter = "GAPDH_expr",
Value = 1,
Description = "Minimum threshold for GAPDH expression"
),
list(
Parameter = "n_cells",
Value = ncol(sce),
Description = "Number of cells in the filtered dataset"
),
list(
Parameter = "n_genes",
Value = nrow(sce),
Description = "Number of genes in the filtered dataset"
),
list(
Parameter = "median_genes",
Value = median(colSums(counts(sce) != 0)),
Description = paste("Median number of expressed genes per cell in the",
"filtered dataset")
)
), pretty = TRUE)
kable(fromJSON(params))
```
Output files
------------
This table describes the output files produced by this document. Right click
and _Save Link As..._ to download the results.
```{r save}
write_rds(sce, here("data/processed/Lindstrom_SCE_filtered.Rds"))
```
```{r output}
dir.create(here("output", DOCNAME), showWarnings = FALSE)
write_lines(params, here("output", DOCNAME, "parameters.json"))
kable(data.frame(
File = c(
glue("[parameters.json]({getDownloadURL('parameters.json', DOCNAME)})")
),
Description = c(
"Parameters set and used in this analysis"
)
))
```