Combine GRanges(List) to a single GRanges with unified breakpoints
1
0
Entering edit mode
Rodrigo • 0
@3e28fb54
Last seen 23 days ago
United States

I posted this question on Stackoverfow: https://stackoverflow.com/questions/72807341/combine-grangeslist-with-unified-breakpoints

I'm looking to merge a list of 6 GRanges objects into one, with the resulting table have a common set of breakpoints across all 6 objects as shown in Figure 1 below. I've tried to do this using findOverlaps(), which projects one coordinate set onto another, without exactly taking both the coordinate sets into account. Most recently, I've experimented with plyranges however, these duplicate the row coordinates rather than breaking them into individual breakpoints, and I'm not exactly sure what's happening with the metadata.

Figure 1: Depiction of genomic segments in three samples (S1..S3), along a chromosome region. The unified breakpoint (UB)ranges are marked by the black vertical lines, labeled 1..9. Figure 1: Depiction of genomic segments in three samples (S1..S3), along a chromosome region. The unified breakpoint (UB) ranges are marked by the black vertical lines, and labeled 1..9.

Any suggestions on how to combine the sets ? I'm open to other suggestions since the data can be coerced to a data.frame, text files, etc.

thank you,

Example using plyranges:

1   > if (identical(getOption('pager'), file.path(R.home('bin'), 'pager'))) options(pager='cat') # rather take the ESS one
2   > options(STERM='iESS', str.dendrogram.last="'", editor='emacsclient', show.error.locations=TRUE)
3   > library(GenomicRanges)
4   > library(tidyverse)
5   > library(plyranges)
6   *** output flushed ***
7   > data(grL)
8   > grL[c(1,6)]
9   GRangesList object of length 2:
10  $A
11  GRanges object with 14212 ranges and 7 metadata columns:
12            seqnames            ranges strand |        Type X.log10.q.value
13               <Rle>         <IRanges>  <Rle> | <character>       <numeric>
14        [1]        1     564424-846680      * |         Amp               0
15        [2]        1     852863-974841      * |         Amp               0
16        [3]        1    986491-1565745      * |         Amp               0
17        [4]        1   1566563-2719525      * |         Amp               0
18        [5]        1   2728545-2761325      * |         Amp               0
19        ...      ...               ...    ... .         ...             ...
20    [14208]       22 50750387-50763701      * |         Del        1.129787
21    [14209]       22 50780464-50885937      * |         Del        2.941190
22    [14210]       22 50887704-51067512      * |         Del        0.011509
23    [14211]       22 51067512-51173133      * |         Del        0.400633
24    [14212]       22 51174234-51178205      * |         Del        0.480761
25              a.score           b.score   c.score      cohort     subtype
26            <numeric>         <numeric> <numeric> <character> <character>
27        [1]  0.163684          1.322534  0.054545    NYC        A
28        [2]  0.137531          1.689206  0.036364    NYC        A
29        [3]  0.064422          0.889654  0.036364    NYC        A
30        [4]  0.092361          1.195796  0.036364    NYC        A
31        [5]  0.101790          0.881394  0.054545    NYC        A
32        ...       ...               ...       ...         ...         ...
33    [14208]  0.068896          0.305572  0.181818    NYC        A
34    [14209]  0.112202          0.305572  0.181818    NYC        A
35    [14210]  0.015748          0.251801  0.181818    NYC        A
36    [14211]  0.037401          0.251801  0.181818    NYC        A
37    [14212]  0.039369          0.278090  0.181818    NYC        A
38    -------
39    seqinfo: 24 sequences from an unspecified genome; no seqlengths
40
41  $F
42  GRanges object with 25101 ranges and 7 metadata columns:
43            seqnames            ranges strand |        Type X.log10.q.value
44               <Rle>         <IRanges>  <Rle> | <character>       <numeric>
45        [1]        1     564424-835336      * |         Amp               0
46        [2]        1    837386-1458139      * |         Amp               0
47        [3]        1   1464470-1555673      * |         Amp               0
48        [4]        1   1556444-1565745      * |         Amp               0
49        [5]        1   1566563-2633368      * |         Amp               0
50        ...      ...               ...    ... .         ...             ...
51    [25097]       22 50710971-50742016      * |         Del        0.926994
52    [25098]       22 50780464-50885937      * |         Del        1.208032
53    [25099]       22 50887704-50914027      * |         Del        0.825840
54    [25100]       22 50915850-51173133      * |         Del        0.844370
55    [25101]       22 51174234-51218950      * |         Del        0.893049
56              a.score           b.score   c.score      cohort     subtype
57            <numeric>         <numeric> <numeric> <character> <character>
58        [1]  0.263871          0.408836  0.269231    NYC        B
59        [2]  0.239878          0.400169  0.250000    NYC        B
60        [3]  0.262476          0.436118  0.250000    NYC        B
61        [4]  0.223937          0.400252  0.230769    NYC        B
62        [5]  0.271884          0.451085  0.250000    NYC        B
63        ...       ...               ...       ...         ...         ...
64    [25097]  0.224858          0.350757  0.346154    NYC        B
65    [25098]  0.247760          0.362135  0.365385    NYC        B
66    [25099]  0.214448          0.338461  0.384615    NYC        B
67    [25100]  0.216530          0.338461  0.384615    NYC        B
68    [25101]  0.220694          0.378862  0.346154    NYC        B
69    -------
70    seqinfo: 24 sequences from an unspecified genome; no seqlengths
71
72  > a <- join_overlap_left(grL[[2]], grL[[1]])
73  > a
74  GRanges object with 73802 ranges and 14 metadata columns:
75            seqnames            ranges strand |      Type.x X.log10.q.value.x
76               <Rle>         <IRanges>  <Rle> | <character>         <numeric>
77        [1]        1     564424-835336      * |         Amp                 0
78        [2]        1     564424-835336      * |         Amp                 0
79        [3]        1    837386-1458139      * |         Amp                 0
80        [4]        1    837386-1458139      * |         Amp                 0
81        [5]        1    837386-1458139      * |         Amp                 0
82        ...      ...               ...    ... .         ...               ...
83    [73798]       22 50915850-51173133      * |         Del          0.844370
84    [73799]       22 50915850-51173133      * |         Del          0.844370
85    [73800]       22 50915850-51173133      * |         Del          0.844370
86    [73801]       22 51174234-51218950      * |         Del          0.893049
87    [73802]       22 51174234-51218950      * |         Del          0.893049
88            a.score.x           b.score.x   c.score.x    cohort.x   subtype.x
89            <numeric>           <numeric>   <numeric> <character> <character>
90        [1]  0.263871            0.408836    0.269231    NYC        B
91        [2]  0.263871            0.408836    0.269231    NYC        B
92        [3]  0.239878            0.400169    0.250000    NYC        B
93        [4]  0.239878            0.400169    0.250000    NYC        B
94        [5]  0.239878            0.400169    0.250000    NYC        B
95        ...       ...                 ...         ...         ...         ...
96    [73798]  0.216530            0.338461    0.384615    NYC        B
97    [73799]  0.216530            0.338461    0.384615    NYC        B
98    [73800]  0.216530            0.338461    0.384615    NYC        B
99    [73801]  0.220694            0.378862    0.346154    NYC        B
100   [73802]  0.220694            0.378862    0.346154    NYC        B
101                Type.y X.log10.q.value.y a.score.y           b.score.y
102           <character>         <numeric> <numeric>           <numeric>
103       [1]         Del          0.006434  0.007874            0.201368
104       [2]         Amp          0.000000  0.163684            1.322534
105       [3]         Del          0.006434  0.007874            0.201368
106       [4]         Amp          0.000000  0.163684            1.322534
107       [5]         Amp          0.000000  0.137531            1.689206
108       ...         ...               ...       ...                 ...
109   [73798]         Amp          0.000000  0.000000            0.000000
110   [73799]         Del          0.011509  0.015748            0.251801
111   [73800]         Del          0.400633  0.037401            0.251801
112   [73801]         Amp          0.000000  0.000000            0.000000
113   [73802]         Del          0.480761  0.039369            0.278090
114             c.score.y    cohort.y   subtype.y
115             <numeric> <character> <character>
116       [1]    0.018182    NYC        A
117       [2]    0.054545    NYC        A
118       [3]    0.018182    NYC        A
119       [4]    0.054545    NYC        A
120       [5]    0.036364    NYC        A
121       ...         ...         ...         ...
122   [73798]    0.000000    NYC        A
123   [73799]    0.181818    NYC        A
124   [73800]    0.181818    NYC        A
125   [73801]    0.000000    NYC        A
126   [73802]    0.181818    NYC        A
127   -------
128   seqinfo: 24 sequences from an unspecified genome; no seqlengths
129
130 > sessionInfo()
131 R version 4.0.5 Patched (2021-06-20 r80951)
132 Platform: x86_64-apple-darwin17.0 (64-bit)
133 Running under: macOS Big Sur 10.16
134
135 Matrix products: default
136 BLAS:   /Library/Frameworks/R.framework/Versions/4.0/Resources/lib/libRblas.0.dylib
137 LAPACK: /Library/Frameworks/R.framework/Versions/4.0/Resources/lib/libRlapack.dylib
138
139 locale:
140 [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
141
142 attached base packages:
143 [1] parallel  stats4    stats     graphics  grDevices utils     datasets
144 [8] methods   base
145
146 other attached packages:
147  [1] plyranges_1.10.0     forcats_0.5.1        stringr_1.4.0
148  [4] dplyr_1.0.8          purrr_0.3.4          readr_2.1.2
149  [7] tidyr_1.2.0          tibble_3.1.6         ggplot2_3.3.6
150 [10] tidyverse_1.3.1      GenomicRanges_1.42.0 GenomeInfoDb_1.26.7
151 [13] IRanges_2.24.1       S4Vectors_0.28.1     BiocGenerics_0.36.1
152
153 loaded via a namespace (and not attached):
154  [1] lattice_0.20-45             lubridate_1.8.0
155  [3] Rsamtools_2.6.0             Biostrings_2.58.0
156  [5] assertthat_0.2.1            rprojroot_2.0.3
157  [7] utf8_1.2.2                  R6_2.5.1
158  [9] cellranger_1.1.0            backports_1.4.1
159 [11] reprex_2.0.1                httr_1.4.3
160 [13] pillar_1.7.0                zlibbioc_1.36.0
161 [15] rlang_1.0.2                 readxl_1.4.0
162 [17] rstudioapi_0.13             Matrix_1.4-1
163 [19] BiocParallel_1.24.1         RCurl_1.98-1.6
164 [21] munsell_0.5.0               DelayedArray_0.16.3
165 [23] broom_0.8.0                 rtracklayer_1.50.0
166 [25] compiler_4.0.5              modelr_0.1.8
167 [27] pkgconfig_2.0.3             tidyselect_1.1.2
168 [29] SummarizedExperiment_1.20.0 GenomeInfoDbData_1.2.4
169 [31] matrixStats_0.61.0          XML_3.99-0.9
170 [33] fansi_1.0.3                 crayon_1.5.1
171 [35] tzdb_0.3.0                  dbplyr_2.1.1
172 [37] withr_2.5.0                 GenomicAlignments_1.26.0
173 [39] bitops_1.0-7                grid_4.0.5
174 [41] jsonlite_1.8.0              gtable_0.3.0
175 [43] lifecycle_1.0.1             DBI_1.1.3
176 [45] magrittr_2.0.3              scales_1.2.0
177 [47] cli_3.2.0                   stringi_1.7.6
178 [49] XVector_0.30.0              fs_1.5.2
179 [51] xml2_1.3.3                  ellipsis_0.3.2
180 [53] generics_0.1.2              vctrs_0.4.0
181 [55] tools_4.0.5                 Biobase_2.50.0
182 [57] glue_1.6.2                  hms_1.1.1
183 [59] MatrixGenerics_1.2.1        colorspace_2.0-3
184 [61] rvest_1.0.2                 haven_2.4.3
185 [63] usethis_2.1.6
186 >
GenomicRanges plyranges • 188 views
ADD COMMENT
2
Entering edit mode
Malcolm Cook ★ 1.6k
@malcolm-cook-6293
Last seen 2 days ago
United States

I find that your problem statement suffers in many ways:

  • it mentions 6 Granges objects but I don't see 6 of anything in the figure
  • the figure has labels S1, S2, S3, UB which are nowhere mentioned elsewhere in your statement or in your example code
  • the function of findOverlaps is mis-stated
  • though you provide the output of printing a putative example, you don't tell us how to create such an example. You need to provide us with the source of data(grl)
  • how to handle the metatdata (mcols) of the component ranges is not defined.

So, it is a bit of a reach on my part, but I think you are looking for GenomicRanges::reduce and that you may find that a step in the direction you are headed

Modifying the Example from the disjoin manpage to set gr1$strand to '-' instead of '+':

     gr1 <- GRanges(seqnames="chr2", ranges=IRanges(3, 6),
                    strand="-", score=5L, GC=0.45)
     gr2 <- GRanges(seqnames="chr1",
                    ranges=IRanges(c(10, 7, 19), width=5),
                    strand=c("+", "-", "+"), score=3:5, GC=c(0.3, 0.5, 0.66))
     gr3 <- GRanges(seqnames=c("chr1", "chr2"),
                    ranges=IRanges(c(1, 4), c(3, 9)),
                    strand=c("-", "-"), score=c(6L, 2L), GC=c(0.4, 0.1))
     grl <- GRangesList(gr1=gr1, gr2=gr2, gr3=gr3)

Let's see what disjoin does:

> ugrl<-unlist(grl)
> ugrl
GRanges object with 6 ranges and 2 metadata columns:
      seqnames    ranges strand |     score        GC
         <Rle> <IRanges>  <Rle> | <integer> <numeric>
  gr1     chr2       3-6      - |         5      0.45
  gr2     chr1     10-14      + |         3      0.30
  gr2     chr1      7-11      - |         4      0.50
  gr2     chr1     19-23      + |         5      0.66
  gr3     chr1       1-3      - |         6      0.40
  gr3     chr2       4-9      - |         2      0.10
  -------
  seqinfo: 2 sequences from an unspecified genome; no seqlengths
> dugrl<-disjoin(ugrl, with.revmap=TRUE)
> dugrl
GRanges object with 7 ranges and 1 metadata column:
      seqnames    ranges strand |        revmap
         <Rle> <IRanges>  <Rle> | <IntegerList>
  [1]     chr2         3      - |             1
  [2]     chr2       4-6      - |           1,6
  [3]     chr2       7-9      - |             6
  [4]     chr1     10-14      + |             2
  [5]     chr1     19-23      + |             4
  [6]     chr1       1-3      - |             5
  [7]     chr1      7-11      - |             3
  -------
  seqinfo: 2 sequences from an unspecified genome; no seqlengths

Note chr 4-6 is only region where there was overlap between component regions (created by my earlier modification to the example data).

I'm guessing this is an example of computing what you are calling the common set of breakpoints.

What the metadata columns of this result should be can be keyed off the revmap according to your needs.

For example, if you want to "carry forward" the meta-data from the overlapping region having the highest score, you could do this;

> mcols(dugrl)<-do.call(rbind,lapply(dugrl$revmap,function(i) mcols(ugrl[i[which.max(ugrl[i]$score)]])))
> dugrl
GRanges object with 7 ranges and 2 metadata columns:
      seqnames    ranges strand |     score        GC
         <Rle> <IRanges>  <Rle> | <integer> <numeric>
  [1]     chr2         3      - |         5      0.45
  [2]     chr2       4-6      - |         5      0.45
  [3]     chr2       7-9      - |         2      0.10
  [4]     chr1     10-14      + |         3      0.30
  [5]     chr1     19-23      + |         5      0.66
  [6]     chr1       1-3      - |         6      0.40
  [7]     chr1      7-11      - |         4      0.50
  -------
  seqinfo: 2 sequences from an unspecified genome; no seqlengths

If you want a better answer we need a better posed question.

ADD COMMENT
0
Entering edit mode

@malcolm-cook-6293, thank you for your response, I'm working through your code as we speak.

ADD REPLY
0
Entering edit mode

Hi @malcolm-cook-6293 , Thanks again for the help. The code got me started albeit, I changed a bit the handling of the metadata.

#' create a list of GRanges with a common set of breakpoints
#' @param grl a granges list object
#' @return
#' A coerced `data.frame` list
#' @export
unified_common_breakpoints <- function(grl) {
    gru <- unlist(grl) %>%
        plyranges::arrange(seqnames, start)

    dgru <- disjoin(gru, with.revmap = TRUE) %>%
        plyranges::arrange(seqnames, start)

    dgru.df <- dgru %>% as.data.frame

    arL <- lapply(grl, function(i) {
        ## find overlaps, use type = "any" (default), and select the first hit
        ## impute NA using zoo::na.approx, and round to nearest integer
        ovr <- findOverlaps(dgru, i, select = "first") %>%
            zoo::na.approx() %>% round()

        ## expand the gistic.scores to new overlaps
        out <- i[ovr, ] %>% as.data.frame %>%
            mutate(orig.chrom = seqnames,
                   orig.start = start,
                   orig.end = end,
                   orig.width = width) %>%
            select(-seqnames, -start, -end, -width, -strand)

        ## add new coordinates
        ## remove X and Y as these cannot be imputed
        out <- cbind(dgru.df[1:length(ovr), 1:3], out)
        return(out)

    })
    return(arL)
}

## run fx()
ubmGS.grL <- unified_common_breakpoints(gistic.scores.grL)
#>                A_W_i              A_W_j              A_D_i              A_D_j
#>                83158              83158              83159              83159
#>                B_W_i              B_W_j              B_D_i              B_D_j
#>                83864              83864              83864              83864
#>                C_W_i              C_W_j              C_D_i              C_D_j
#>                83948              83948              83948              83948

ubmGSL <- do.call(rbind, ubmGS.grL)
ubmGS <- ubmGSL %>%
    pivot_wider(id_cols = c("seqnames", "start", "end", "Type"),
                values_from = "score.b",
                names_from = c("cohort", "subtype"))

head(ubmGS)
#>      # A tibble: 6 × 10
#>        seqnames  start    end Type          A_W           A_D         B_W
#>        <fct>     <int>  <int> <chr>       <dbl>         <dbl>       <dbl>
#>      1 1             0 564423 i               0             0           0
#>      2 1        564424 835336 i               0             0           0
#>      3 1        835337 837385 i               0             0           0
#>      4 1        837386 846680 i               0             0           0
#>      5 1        846681 852862 i               0             0           0
#>      6 1        852863 974841 i               0             0           0
#>      # … with 3 more variables: B_D <dbl>, C_W <dbl>, C_D <dbl>
ADD REPLY

Login before adding your answer.

Traffic: 205 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6