Calculate and segment scores for footprints and add them to a SummarizedExeriment

Given a read-level footprinting data and a weight vector defining a specific footprint, calculate footprinting scores, identify high-scoring segments and add them to the input SummarizedExeriment object.

Typicallly, this can be done using addFootprints which will:

Calculate footprint scores for each individual read using calcFootprintScores, given a SummarizedExperiment with a read-level assay and a vector of weights that defines the footprint (see wgt parameter).
Segment footprint scores to identify high-scoring segments in the scores calculated at step 1 using segmentFootprintScores.
Store high-scoring segments in the SummarizedExperiment as a column in colData of the object.

If required, the functions calcFootprintScores or segmentFootprintScores can also be called directly to perform only one of the steps above.

The footprint is defined by a weight vector wgt which describes its length in bases and the shape of the expected modification profile. The score is based on a cross-correlation of the modification probabilities (centered by subtracting 0.5) with wgt, weighted by the minimum of minweight and the modification probability.

Usage

addFootprints(
  se,
  wgt,
  assayName = "mod_prob",
  minconf = 0.7,
  minweight = 0.05,
  minperiod = 50,
  maxperiod = 300,
  thresh = 0.15,
  lenRange = c(10, 80),
  width = 140,
  name = "footprints",
  verbose = TRUE
)

calcFootprintScores(
  se,
  wgt,
  assayName = "mod_prob",
  minconf = 0.7,
  minweight = 0.05,
  verbose = TRUE
)

segmentFootprintScores(
  scoresList,
  minperiod = 50,
  maxperiod = 300,
  thresh = 0.15,
  lenRange = c(10, 80),
  width = 140,
  verbose = TRUE
)

Arguments

se: A SummarizedExperiment object.
wgt: Numeric vector with weights that define the footprint to score. Typically centered at zero.
assayName: A character scalar providing the name of a read-level assay in se that contains modification probabilities.
minconf: Numeric scalar giving the minimal confidence of a base to be included in the calculation of footprint scores.
minweight: Numeric scalar giving the minimal weight for position in the calculation of footprint scores.
minperiod, maxperiod: Numeric scalars that define the band-pass filter parameters used to smooth the footprint scores for segmentation. minperiod and maxperiod give the minimal and maximal periods (in base pairs) for the critical frequencies in the score signal that should be retained. The default values are suitable to retain nucleosomal signals. The filtering is performed using filtfilt and thus requires the signal package to be installed.
thresh: A numeric scalar giving the minimal smoothed score of a footprint. Higher values make the footprint detection more stringent.
lenRange: A numeric vector with two elements giving the minimal and maximal number of consecutive score values (base pairs) that must be greater than thresh for a region to be included in the returned footprints.
width: A numeric scalar giving the width of returned footprints. Footprints are created by placing a region of length width centered on the maximal smoothed score value in each identified high-scoring segment.
name: Character scalar giving the column name in colData(se) in which the high-scoring footprints are stored as a list (over samples) or IRangesLists (over reads).
verbose: If TRUE, report on progress.
scoresList: A list of data.frames with footprint scores as returned by calcFootprintScores.

Value

For addFootprints: , a SummarizedExperiment object with footprints in colData(se)[[name]].
For calcFootprintScores: , a list with one entry per sample (column) in se. The list elements are data.frames with columns, readId, pos, pmod and score, in (continuous) base space from min(pos(rowRanges(se))) to max(pos(rowRanges(se))) (only including positions covered by reads in a given sample).
For segmentFootprintScores: , a list (over samples) of IRangesList elements (over reads). Each element contains one IRanges object with the identified footprints of a single read (all of length with bases).

Author

Michael Stadler, Charlotte Soneson

Examples

# load example data
modbamfile <- system.file("extdata", "6mA_1_10reads.bam",
                          package = "footprintR")
se <- readModBam(bamfiles = modbamfile, regions = "chr1:6940000-6955000",
                 modbase = "a", verbose = FALSE,
                 BPPARAM = BiocParallel::SerialParam())

# add nucleosome footprints
# "wgt": footprint weights for nucleosome with flanks
# - weights are centered around zero (starting with -0.5 and 0.5)
# - weights are adjusted for their relative frequency, to give similar
#   importance to flanks and protected region
# - weights are repeated for bases in the (un)modified parts (15 bases in
#   flanks and 140 bases protected by the nucleosome)
wgt <- rep(c(0.5, -0.5, 0.5) * c(140/170, 30/170, 140/170), c(15, 140, 15))
se <- addFootprints(se, wgt, thresh = 0.05, name = "nucl")
#> ℹ calculating footprint scores
#> ✔ calculating footprint scores [34ms]
#> 
#> ⠙ 0.0 thousand records processed (16.6 /s) [61ms]

#> ℹ segmenting footprint scores
#> ✔ segmenting footprint scores [42ms]
#> 
#> ⠙ 0.0 thousand records processed (16.6 /s) [61ms]

se$nucl
#> $s1
#> IRangesList object of length 3:
#> $`s1-233e48a7-f379-4dcf-9270-958231125563`
#> IRanges object with 15 ranges and 0 metadata columns:
#>            start       end     width
#>        <integer> <integer> <integer>
#>    [1]   6926371   6926510       140
#>    [2]   6926557   6926696       140
#>    [3]   6926737   6926876       140
#>    [4]   6926917   6927056       140
#>    [5]   6927105   6927244       140
#>    ...       ...       ...       ...
#>   [11]   6935854   6935993       140
#>   [12]   6936034   6936173       140
#>   [13]   6938096   6938235       140
#>   [14]   6939136   6939275       140
#>   [15]   6940416   6940555       140
#> 
#> $`s1-d52a5f6a-a60a-4f85-913e-eada84bfbfb9`
#> IRanges object with 1 range and 0 metadata columns:
#>           start       end     width
#>       <integer> <integer> <integer>
#>   [1]   6933059   6933198       140
#> 
#> $`s1-92e906ae-cddb-4347-a114-bf9137761a8d`
#> IRanges object with 2 ranges and 0 metadata columns:
#>           start       end     width
#>       <integer> <integer> <integer>
#>   [1]   6931342   6931481       140
#>   [2]   6932761   6932900       140
#> 
#> 

# only calculate score
scoresList <- calcFootprintScores(se, wgt)
#> ℹ calculating footprint scores
#> ✔ calculating footprint scores [30ms]
#> 
#> ⠙ 0.0 thousand records processed (16.6 /s) [61ms]

lapply(scoresList, nrow)
#> $s1
#> [1] 38549
#> 
lapply(scoresList, head)
#> $s1
#>                                    readId     pos pmod score
#> 1 s1-233e48a7-f379-4dcf-9270-958231125563 6925830 -0.5    NA
#> 2 s1-233e48a7-f379-4dcf-9270-958231125563 6925831   NA    NA
#> 3 s1-233e48a7-f379-4dcf-9270-958231125563 6925832   NA    NA
#> 4 s1-233e48a7-f379-4dcf-9270-958231125563 6925833   NA    NA
#> 5 s1-233e48a7-f379-4dcf-9270-958231125563 6925834 -0.5    NA
#> 6 s1-233e48a7-f379-4dcf-9270-958231125563 6925835   NA    NA
#> 

# segment calculated scores
irl <- segmentFootprintScores(scoresList, thresh = 0.05)
#> ℹ segmenting footprint scores
#> ✔ segmenting footprint scores [34ms]
#> 
#> ⠙ 0.0 thousand records processed (16.6 /s) [61ms]

irl$s1
#> IRangesList object of length 3:
#> $`s1-233e48a7-f379-4dcf-9270-958231125563`
#> IRanges object with 15 ranges and 0 metadata columns:
#>            start       end     width
#>        <integer> <integer> <integer>
#>    [1]   6926371   6926510       140
#>    [2]   6926557   6926696       140
#>    [3]   6926737   6926876       140
#>    [4]   6926917   6927056       140
#>    [5]   6927105   6927244       140
#>    ...       ...       ...       ...
#>   [11]   6935854   6935993       140
#>   [12]   6936034   6936173       140
#>   [13]   6938096   6938235       140
#>   [14]   6939136   6939275       140
#>   [15]   6940416   6940555       140
#> 
#> $`s1-d52a5f6a-a60a-4f85-913e-eada84bfbfb9`
#> IRanges object with 1 range and 0 metadata columns:
#>           start       end     width
#>       <integer> <integer> <integer>
#>   [1]   6933059   6933198       140
#> 
#> $`s1-92e906ae-cddb-4347-a114-bf9137761a8d`
#> IRanges object with 2 ranges and 0 metadata columns:
#>           start       end     width
#>       <integer> <integer> <integer>
#>   [1]   6931342   6931481       140
#>   [2]   6932761   6932900       140
#>