-
Notifications
You must be signed in to change notification settings - Fork 0
/
1_mergingSamples.Rmd
361 lines (304 loc) · 18.4 KB
/
1_mergingSamples.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
---
title: "scATAC-seq Rat Metrial Glands"
author: "Ha T. H. Vu"
output: html_document
---
```{r setup, include=FALSE}
options(max.print = "75")
knitr::opts_chunk$set(
echo = TRUE,
collapse = TRUE,
comment = "#>",
fig.path = "Files/",
fig.width = 15,
prompt = FALSE,
tidy = FALSE,
message = FALSE,
warning = TRUE
)
knitr::opts_knit$set(width = 75)
```
This is a documentation for analyses of scATAC-seq data, generated from rat metrial gland tissues on gestational day (GD) 15.5 and 19.5. <br>
We first set up the environment with necessary libraries and functions.
```{r}
library(Signac)
library(Seurat)
library(S4Vectors)
library(GenomicRanges)
library(patchwork)
library(biomaRt)
library(ggplot2)
plotComparisonDf <- function(comparison_df, title){
ggplot(comparison_df, aes(value)) +
geom_histogram(bins = 70, aes(fill=method), color='black') +
theme_classic() +
scale_fill_manual(values=c("TF"="red", "log TF"="#d3d3d3")) +
xlab('value in matrix') +
facet_wrap(~method, scales='free') + ggtitle(title)
}
```
## 1. GD 15.5:
We first filter the low-quality cells, as determined in the previous step.
```{r}
set.seed(1234)
load("/work/LAS/geetu-lab/hhvu/project3_scATAC/scATAC-seq-analysis/1_preprocess/2_MACS2Peaks/3_unfilteredObjects-MACSpeaks/GD15.5_1.rda")
gd15.5_1 <- gd15.5_1_MACS
load("/work/LAS/geetu-lab/hhvu/project3_scATAC/scATAC-seq-analysis/1_preprocess/2_MACS2Peaks/3_unfilteredObjects-MACSpeaks/GD15.5_2.rda")
gd15.5_2 <- gd15.5_2_MACS
load("/work/LAS/geetu-lab/hhvu/project3_scATAC/scATAC-seq-analysis/1_preprocess/2_MACS2Peaks/3_unfilteredObjects-MACSpeaks/GD15.5_3.rda")
gd15.5_3 <- gd15.5_3_MACS
gd15.5_1 <- subset(x = gd15.5_1, subset = peak_region_fragments > 1000 & peak_region_fragments < 20000 & pct_reads_in_peaks > 15 & TSS.enrichment > 1.5)
gd15.5_2 <- subset(x = gd15.5_2, subset = peak_region_fragments > 1000 & peak_region_fragments < 20000 & pct_reads_in_peaks > 15 & TSS.enrichment > 1.5)
gd15.5_3 <- subset(x = gd15.5_3, subset = peak_region_fragments > 1000 & peak_region_fragments < 20000 & pct_reads_in_peaks > 15 & TSS.enrichment > 1.5)
```
Our data shows a trend of skewness; therefore, we should use `method = 3` when running dimension reduction. Codes here were adapted from http://andrewjohnhill.com/blog/2019/05/06/dimensionality-reduction-for-scatac-data/
```{r}
counts <- gd15.5_1@assays$MACSpeaks@counts
tf = t(t(counts) / Matrix::colSums(counts))
comparison_df = rbind(data.frame(value=tf@x, method='TF'), data.frame(value=log1p(tf@x * 100000), method='log TF')) #100000 is scale factor, in Seurat it is default as 10000
plotComparisonDf(comparison_df, "gd 15.5_1")
counts <- gd15.5_2@assays$MACSpeaks@counts
tf = t(t(counts) / Matrix::colSums(counts))
comparison_df = rbind(data.frame(value=tf@x, method='TF'), data.frame(value=log1p(tf@x * 100000), method='log TF')) #100000 is scale factor, in Seurat it is default as 10000
plotComparisonDf(comparison_df, "gd 15.5_2")
counts <- gd15.5_3@assays$MACSpeaks@counts
tf = t(t(counts) / Matrix::colSums(counts))
comparison_df = rbind(data.frame(value=tf@x, method='TF'), data.frame(value=log1p(tf@x * 100000), method='log TF')) #100000 is scale factor, in Seurat it is default as 10000
plotComparisonDf(comparison_df, "gd 15.5_3")
```
Next, we run TF-IDF with `method = 3`.
```{r}
#run TF-IDF
gd15.5_1 <- RunTFIDF(gd15.5_1, assay = 'MACSpeaks', method = 3)
gd15.5_1 <- FindTopFeatures(gd15.5_1, min.cutoff = 'q0')
gd15.5_1 <- RunSVD(gd15.5_1, assay = 'MACSpeaks', reduction.key = 'LSI_', reduction.name = 'lsi')
gd15.5_2 <- RunTFIDF(gd15.5_2, assay = 'MACSpeaks', method = 3)
gd15.5_2 <- FindTopFeatures(gd15.5_2, min.cutoff = 'q0')
gd15.5_2 <- RunSVD(gd15.5_2, assay = 'MACSpeaks', reduction.key = 'LSI_', reduction.name = 'lsi')
gd15.5_3 <- RunTFIDF(gd15.5_3, assay = 'MACSpeaks', method = 3)
gd15.5_3 <- FindTopFeatures(gd15.5_3, min.cutoff = 'q0')
gd15.5_3 <- RunSVD(gd15.5_3, assay = 'MACSpeaks', reduction.key = 'LSI_', reduction.name = 'lsi')
```
We then look at the correlation between the LSI components and the sequencing depth, and the elbow plots, so that we can determine the number of components to keep for clustering.
```{r}
DepthCor(gd15.5_1, n=50) + ggtitle("gd 15.5_1")
DepthCor(gd15.5_2, n=50) + ggtitle("gd 15.5_2")
DepthCor(gd15.5_3, n=50) + ggtitle("gd 15.5_3")
ElbowPlot(gd15.5_1, n=50, reduction="lsi") + ggtitle("gd 15.5_1")
ElbowPlot(gd15.5_2, n=50, reduction="lsi") + ggtitle("gd 15.5_2")
ElbowPlot(gd15.5_3, n=50, reduction="lsi") + ggtitle("gd 15.5_3")
```
From the plots, we will keep the first 20 components, then exclude component 1 because of its high correlation with sequencing depth.
```{r, eval=F}
gd15.5_1 <- RunUMAP(object = gd15.5_1, reduction = 'lsi', dims = 2:20) #drop the 1st component, see DepthCor plot and elbow plot
gd15.5_2 <- RunUMAP(object = gd15.5_2, reduction = 'lsi', dims = 2:20)
gd15.5_3 <- RunUMAP(object = gd15.5_3, reduction = 'lsi', dims = 2:20)
```
In order to merge the replicates, we need to find common peaks across replicates. We obtain this set of peaks here.
```{r, eval=F}
intersecting.regions_15.5_2_3 <- findOverlaps(query = gd15.5_2, subject = gd15.5_3)
intersections.15.5_2 <- unique(queryHits(intersecting.regions_15.5_2_3))
peaks.15.5_2_3 <- sort(granges(gd15.5_2)[intersections.15.5_2])
intersecting.regions_15.5_1_2_3 <- findOverlaps(query=gd15.5_1, subject=peaks.15.5_2_3)
intersections.15.5_1 <- unique(queryHits(intersecting.regions_15.5_1_2_3))
combined.peaks <- sort(granges(gd15.5_1)[intersections.15.5_1])
```
Next, we build a new Seurat object with the new peaks. It will require cell identities and fragment counts from each individual replicate.
```{r, eval=F}
#load original cell identities
md.15.5_1 <- read.table(file = "/work/LAS/geetu-lab/hhvu/project3_scATAC/data/scATAC-seq/GD15.5/ATAC-15.5-1/singlecell.csv",
stringsAsFactors = FALSE, sep = ",", header = TRUE, row.names = 1)[-1, ]
md.15.5_2 <- read.table(file = "/work/LAS/geetu-lab/hhvu/project3_scATAC/data/scATAC-seq/GD15.5/ATAC-15.5-2/singlecell.csv",
stringsAsFactors = FALSE, sep = ",", header = TRUE, row.names = 1)[-1, ]
md.15.5_3 <- read.table(file = "/work/LAS/geetu-lab/hhvu/project3_scATAC/data/scATAC-seq/GD15.5/ATAC-15.5-3/singlecell.csv",
stringsAsFactors = FALSE, sep = ",", header = TRUE, row.names = 1)[-1, ]
#load original fragment counts
frags.15.5_1 <- CreateFragmentObject(path = "/work/LAS/geetu-lab/hhvu/project3_scATAC/data/scATAC-seq/GD15.5/ATAC-15.5-1/fragments.tsv.gz",
cells = rownames(md.15.5_1))
frags.15.5_2 <- CreateFragmentObject(path = "/work/LAS/geetu-lab/hhvu/project3_scATAC/data/scATAC-seq/GD15.5/ATAC-15.5-2/fragments.tsv.gz",
cells = rownames(md.15.5_2))
frags.15.5_3 <- CreateFragmentObject(path = "/work/LAS/geetu-lab/hhvu/project3_scATAC/data/scATAC-seq/GD15.5/ATAC-15.5-3/fragments.tsv.gz",
cells = rownames(md.15.5_3))
#count fragments in the common peaks among replicates
gd15.5_1_peaks <- FeatureMatrix(fragments=frags.15.5_1, features=combined.peaks, cells=colnames(gd15.5_1))
gd15.5_2_peaks <- FeatureMatrix(fragments=frags.15.5_2, features=combined.peaks, cells=colnames(gd15.5_2))
gd15.5_3_peaks <- FeatureMatrix(fragments=frags.15.5_3, features=combined.peaks, cells=colnames(gd15.5_3))
#build new assays to associate with the merged object
gd15.5_1_assay <- CreateChromatinAssay(counts = gd15.5_1_peaks, min.cells = 1, ranges = combined.peaks)
gd15.5_2_assay <- CreateChromatinAssay(counts = gd15.5_2_peaks, min.cells = 1, ranges = combined.peaks)
gd15.5_3_assay <- CreateChromatinAssay(counts = gd15.5_3_peaks, min.cells = 1, ranges = combined.peaks)
gd15.5_1[["combined"]] <- gd15.5_1_assay
gd15.5_2[["combined"]] <- gd15.5_2_assay
gd15.5_3[["combined"]] <- gd15.5_3_assay
DefaultAssay(gd15.5_1) <- "combined"
gd15.5_1 <- RunTFIDF(gd15.5_1, method=3)
gd15.5_1$dataset <- 'gd15.5_1'
DefaultAssay(gd15.5_2) <- "combined"
gd15.5_2 <- RunTFIDF(gd15.5_2, method=3)
gd15.5_2$dataset <- 'gd15.5_2'
DefaultAssay(gd15.5_3) <- "combined"
gd15.5_3 <- RunTFIDF(gd15.5_3, method=3)
gd15.5_3$dataset <- 'gd15.5_3'
#merge data
gd15.5combined <- merge(x = gd15.5_1, y = list(gd15.5_2, gd15.5_3), add.cell.ids = c("15.5_1", "15.5_2", "15.5_3"))
DefaultAssay(gd15.5combined) <- "combined"
gd15.5combined <- RunTFIDF(gd15.5combined, method=3)
gd15.5combined <- FindTopFeatures(gd15.5combined, min.cutoff = 'q0')
gd15.5combined <- RunSVD(gd15.5combined)
gd15.5combined <- RunUMAP(gd15.5combined, reduction = 'lsi', dims = 2:20)
```
With the new combined dataset, we need new fragment counts that are combined from the replicates. In BASH, carry out the following commands.
```
cd /work/LAS/geetu-lab/hhvu/project3_scATAC/data/scATAC-seq/GD15.5
mkdir integratedFragments
for i in {1..3}; do gzip -d ATAC-15.5-"$i"/fragments.tsv.gz; cp ATAC-15.5-"$i"/fragments.tsv integratedFragments/atac15.5_"$i"_fragments.tsv; c=`echo 15.5_"$i"_`; awk -v c=$c '{print $1"\t"$2"\t"$3"\t"c$4"\t"$5}' integratedFragments/atac15.5_"$i"_fragments.tsv > temp; mv temp integratedFragments/atac15.5_"$i"_fragments.tsv; done
cd integratedFragments
ml samtools
sort -m -k 1,1V -k2,2n atac15.5_1_fragments.tsv atac15.5_2_fragments.tsv atac15.5_3_fragments.tsv > fragments.tsv
bgzip -@ 4 fragments.tsv
tabix --preset=bed fragments.tsv.gz
```
```{r, eval=F}
#build new Fragments() for the merge data
frag.path <- "/work/LAS/geetu-lab/hhvu/project3_scATAC/data/scATAC-seq/GD15.5/integratedFragments/fragments.tsv.gz"
fragments <- CreateFragmentObject(path = frag.path, cells = colnames(gd15.5combined), validate.fragments = TRUE)
Fragments(gd15.5combined) <- fragments
#cluster on the new merge data
gd15.5combined <- FindNeighbors(object = gd15.5combined, reduction = 'lsi', dims = 2:20)
gd15.5combined <- FindClusters(object = gd15.5combined, algorithm = 3, resolution = 0.8, verbose = FALSE)
#save(gd15.5combined, file="/work/LAS/geetu-lab/hhvu/project3_scATAC/scATAC-seq-analysis/2_mergedSamples/2_MACS2Peaks/15.5_1-2-3-merged.rda")
```
## 2. GD 19.5:
The analyses in this section is similar to those for gd 15.5 We first filter the low-quality cells.
```{r}
set.seed(1234)
load("/work/LAS/geetu-lab/hhvu/project3_scATAC/scATAC-seq-analysis/1_preprocess/2_MACS2Peaks/3_unfilteredObjects-MACSpeaks/GD19.5_4.rda")
gd19.5_4 <- gd19.5_4_MACS
load("/work/LAS/geetu-lab/hhvu/project3_scATAC/scATAC-seq-analysis/1_preprocess/2_MACS2Peaks/3_unfilteredObjects-MACSpeaks/GD19.5_5.rda")
gd19.5_5 <- gd19.5_5_MACS
load("/work/LAS/geetu-lab/hhvu/project3_scATAC/scATAC-seq-analysis/1_preprocess/2_MACS2Peaks/3_unfilteredObjects-MACSpeaks/GD19.5_6.rda")
gd19.5_6 <- gd19.5_6_MACS
gd19.5_4 <- subset(x = gd19.5_4, subset = peak_region_fragments > 1000 & peak_region_fragments < 20000 & pct_reads_in_peaks > 15 & TSS.enrichment > 1.5)
gd19.5_5 <- subset(x = gd19.5_5, subset = peak_region_fragments > 1000 & peak_region_fragments < 20000 & pct_reads_in_peaks > 15 & TSS.enrichment > 1.5)
gd19.5_6 <- subset(x = gd19.5_6, subset = peak_region_fragments > 1000 & peak_region_fragments < 20000 & pct_reads_in_peaks > 15 & TSS.enrichment > 1.5)
```
Our data shows a trend of skewness in the gd19.5 samples as well.
```{r}
counts <- gd19.5_4@assays$MACSpeaks@counts
tf = t(t(counts) / Matrix::colSums(counts))
comparison_df = rbind(data.frame(value=tf@x, method='TF'), data.frame(value=log1p(tf@x * 100000), method='log TF')) #100000 is scale factor, in Seurat it is default as 10000
plotComparisonDf(comparison_df, "gd 19.5_4")
counts <- gd19.5_5@assays$MACSpeaks@counts
tf = t(t(counts) / Matrix::colSums(counts))
comparison_df = rbind(data.frame(value=tf@x, method='TF'), data.frame(value=log1p(tf@x * 100000), method='log TF')) #100000 is scale factor, in Seurat it is default as 10000
plotComparisonDf(comparison_df, "gd 19.5_5")
counts <- gd19.5_6@assays$MACSpeaks@counts
tf = t(t(counts) / Matrix::colSums(counts))
comparison_df = rbind(data.frame(value=tf@x, method='TF'), data.frame(value=log1p(tf@x * 100000), method='log TF')) #100000 is scale factor, in Seurat it is default as 10000
plotComparisonDf(comparison_df, "gd 19.5_6")
```
Next, we run TF-IDF with `method = 3`.
```{r}
#run TF-IDF
gd19.5_4 <- RunTFIDF(gd19.5_4, assay = 'MACSpeaks', method = 3)
gd19.5_4 <- FindTopFeatures(gd19.5_4, min.cutoff = 'q0')
gd19.5_4 <- RunSVD(gd19.5_4, assay = 'MACSpeaks', reduction.key = 'LSI_', reduction.name = 'lsi')
gd19.5_5 <- RunTFIDF(gd19.5_5, assay = 'MACSpeaks', method = 3)
gd19.5_5 <- FindTopFeatures(gd19.5_5, min.cutoff = 'q0')
gd19.5_5 <- RunSVD(gd19.5_5, assay = 'MACSpeaks', reduction.key = 'LSI_', reduction.name = 'lsi')
gd19.5_6 <- RunTFIDF(gd19.5_6, assay = 'MACSpeaks', method = 3)
gd19.5_6 <- FindTopFeatures(gd19.5_6, min.cutoff = 'q0')
gd19.5_6 <- RunSVD(gd19.5_6, assay = 'MACSpeaks', reduction.key = 'LSI_', reduction.name = 'lsi')
```
We then look at the correlation between the LSI components and the sequencing depth, and the elbow plots, so that we can determine the number of components to keep for clustering.
```{r}
DepthCor(gd19.5_4, n=50) + ggtitle("gd 19.5_4")
DepthCor(gd19.5_5, n=50) + ggtitle("gd 19.5_5")
DepthCor(gd19.5_6, n=50) + ggtitle("gd 19.5_6")
ElbowPlot(gd19.5_4, n=50, reduction="lsi") + ggtitle("gd 19.5_4")
ElbowPlot(gd19.5_5, n=50, reduction="lsi") + ggtitle("gd 19.5_5")
ElbowPlot(gd19.5_6, n=50, reduction="lsi") + ggtitle("gd 19.5_6")
```
From the plots, we will keep the first 10 components, then exclude component 1 because of its high correlation with sequencing depth.
```{r, eval=F}
gd19.5_4 <- RunUMAP(object = gd19.5_4, reduction = 'lsi', dims = 2:10) #drop the 1st component, see DepthCor plot and elbow plot
gd19.5_5 <- RunUMAP(object = gd19.5_5, reduction = 'lsi', dims = 2:10)
gd19.5_6 <- RunUMAP(object = gd19.5_6, reduction = 'lsi', dims = 2:10)
```
In order to merge the replicates, we need to find common peaks across replicates. We obtain this set of peaks here.
```{r, eval=F}
intersecting.regions_19.5_5_6 <- findOverlaps(query = gd19.5_5, subject = gd19.5_6)
intersections.19.5_5 <- unique(queryHits(intersecting.regions_19.5_5_6))
peaks.19.5_5_6 <- sort(granges(gd19.5_5)[intersections.19.5_5])
intersecting.regions_19.5_4_5_6 <- findOverlaps(query=gd19.5_4, subject=peaks.19.5_5_6)
intersections.19.5_4 <- unique(queryHits(intersecting.regions_19.5_4_5_6))
combined.peaks <- sort(granges(gd19.5_4)[intersections.19.5_4])
```
Next, we build a new Seurat object with the new peaks. It will require cell identities and fragment counts from each individual replicate.
```{r, eval=F}
#load original cell identities
md.19.5_4 <- read.table(file = "/work/LAS/geetu-lab/hhvu/project3_scATAC/data/scATAC-seq/GD19.5/ATAC-19.5-4/singlecell.csv",
stringsAsFactors = FALSE, sep = ",", header = TRUE, row.names = 1)[-1, ]
md.19.5_5 <- read.table(file = "/work/LAS/geetu-lab/hhvu/project3_scATAC/data/scATAC-seq/GD19.5/ATAC-19.5-5/singlecell.csv",
stringsAsFactors = FALSE, sep = ",", header = TRUE, row.names = 1)[-1, ]
md.19.5_6 <- read.table(file = "/work/LAS/geetu-lab/hhvu/project3_scATAC/data/scATAC-seq/GD19.5/ATAC-19.5-6/singlecell.csv",
stringsAsFactors = FALSE, sep = ",", header = TRUE, row.names = 1)[-1, ]
#load original fragment counts
frags.19.5_4 <- CreateFragmentObject(path = "/work/LAS/geetu-lab/hhvu/project3_scATAC/data/scATAC-seq/GD19.5/ATAC-19.5-4/fragments.tsv.gz",
cells = rownames(md.19.5_4))
frags.19.5_5 <- CreateFragmentObject(path = "/work/LAS/geetu-lab/hhvu/project3_scATAC/data/scATAC-seq/GD19.5/ATAC-19.5-5/fragments.tsv.gz",
cells = rownames(md.19.5_5))
frags.19.5_6 <- CreateFragmentObject(path = "/work/LAS/geetu-lab/hhvu/project3_scATAC/data/scATAC-seq/GD19.5/ATAC-19.5-6/fragments.tsv.gz",
cells = rownames(md.19.5_6))
#count fragments in the common peaks among replicates
gd19.5_4_peaks <- FeatureMatrix(fragments= frags.19.5_4, features=combined.peaks, cells=colnames(gd19.5_4))
gd19.5_5_peaks <- FeatureMatrix(fragments= frags.19.5_5, features=combined.peaks, cells=colnames(gd19.5_5))
gd19.5_6_peaks <- FeatureMatrix(fragments= frags.19.5_6, features=combined.peaks, cells=colnames(gd19.5_6))
#build new assays to associate with the merged object
gd19.5_4_assay <- CreateChromatinAssay(counts = gd19.5_4_peaks, min.cells = 1, ranges = combined.peaks)
gd19.5_5_assay <- CreateChromatinAssay(counts = gd19.5_5_peaks, min.cells = 1, ranges = combined.peaks)
gd19.5_6_assay <- CreateChromatinAssay(counts = gd19.5_6_peaks, min.cells = 1, ranges = combined.peaks)
gd19.5_4[["combined"]] <- gd19.5_4_assay
gd19.5_5[["combined"]] <- gd19.5_5_assay
gd19.5_6[["combined"]] <- gd19.5_6_assay
DefaultAssay(gd19.5_4) <- "combined"
gd19.5_4 <- RunTFIDF(gd19.5_4, method=3)
gd19.5_4$dataset <- 'gd19.5_4'
DefaultAssay(gd19.5_5) <- "combined"
gd19.5_5 <- RunTFIDF(gd19.5_5, method=3)
gd19.5_5$dataset <- 'gd19.5_5'
DefaultAssay(gd19.5_6) <- "combined"
gd19.5_6 <- RunTFIDF(gd19.5_6, method=3)
gd19.5_6$dataset <- 'gd19.5_6'
#merge data
gd19.5combined <- merge(x = gd19.5_4, y = list(gd19.5_5, gd19.5_6), add.cell.ids = c("19.5_4", "19.5_5", "19.5_6"))
DefaultAssay(gd19.5combined) <- "combined"
gd19.5combined <- RunTFIDF(gd19.5combined, method=3)
gd19.5combined <- FindTopFeatures(gd19.5combined, min.cutoff = 'q0')
gd19.5combined <- RunSVD(gd19.5combined)
gd19.5combined <- RunUMAP(gd19.5combined, reduction = 'lsi', dims = 2:10)
```
With the new combined dataset, we need new fragment counts that are combined from the replicates. In BASH, carry out the following commands.
```
cd /work/LAS/geetu-lab/hhvu/project3_scATAC/data/scATAC-seq/GD19.5
mkdir integratedFragments
for i in {4..6}; do gzip -d ATAC-19.5-"$i"/fragments.tsv.gz; cp ATAC-19.5-"$i"/fragments.tsv integratedFragments/atac19.5_"$i"_fragments.tsv; c=`echo 19.5_"$i"_`; awk -v c=$c '{print $1"\t"$2"\t"$3"\t"c$4"\t"$5}' integratedFragments/atac19.5_"$i"_fragments.tsv > temp; mv temp integratedFragments/atac19.5_"$i"_fragments.tsv; done
cd integratedFragments
ml samtools
sort -m -k 1,1V -k2,2n atac19.5_4_fragments.tsv atac19.5_5_fragments.tsv atac19.5_6_fragments.tsv > fragments.tsv
bgzip -@ 14 fragments.tsv
tabix --preset=bed fragments.tsv.gz
```
```{r, eval=F}
#build new Fragments() for the merge data
frag.path <- "/work/LAS/geetu-lab/hhvu/project3_scATAC/data/scATAC-seq/GD19.5/integratedFragments/fragments.tsv.gz"
fragments <- CreateFragmentObject(path = frag.path, cells = colnames(gd19.5combined), validate.fragments = TRUE)
Fragments(gd19.5combined) <- fragments
#cluster on the new merge data
gd19.5combined <- FindNeighbors(object = gd19.5combined, reduction = 'lsi', dims = 2:10)
gd19.5combined <- FindClusters(object = gd19.5combined, algorithm = 3, resolution = 0.8, verbose = FALSE)
#save(gd19.5combined, file="/work/LAS/geetu-lab/hhvu/project3_scATAC/scATAC-seq-analysis/2_mergedSamples/2_MACS2Peaks/19.5_4-5-6-merged.rda")
```
```{r}
sessionInfo()
```