In order to find a threshold value to filter lowly expressed features, SeqGate analyzes the distribution of counts found in replicates along with zero counts. More specifically, features with a customizable minimal proportion of zeros in one condition are selected. The distribution of counts found in replicates of that same condition along with those zeros is computed. The chosen threshold is the count value corresponding to the customizable percentile of this distribution. Finally, features having a customizable proportion (90% by default) of replicates with counts below that value in all conditions are filtered. Default value for all customizable parameters have been set through extensive simulation batch testing and can be considered as adequate in most situations.
To install SeqGate, start R and enter:
if (!requireNamespace("BiocManager", quietly = TRUE))
install.packages("BiocManager")
BiocManager::install("SeqGate")## Bioconductor version 3.20 (BiocManager 1.30.25), R 4.4.1 (2024-06-14)## Warning: package(s) not installed when version(s) same as or greater than current; use
## `force = TRUE` to re-install: 'SeqGate'## Old packages: 'fs', 'boot', 'foreign', 'lattice', 'nlme'First load SeqGate:
## Loading required package: S4Vectors## Loading required package: stats4## Loading required package: BiocGenerics##
## Attaching package: 'BiocGenerics'## The following objects are masked from 'package:stats':
##
## IQR, mad, sd, var, xtabs## The following objects are masked from 'package:base':
##
## Filter, Find, Map, Position, Reduce, anyDuplicated, aperm, append,
## as.data.frame, basename, cbind, colnames, dirname, do.call,
## duplicated, eval, evalq, get, grep, grepl, intersect, is.unsorted,
## lapply, mapply, match, mget, order, paste, pmax, pmax.int, pmin,
## pmin.int, rank, rbind, rownames, sapply, saveRDS, setdiff, table,
## tapply, union, unique, unsplit, which.max, which.min##
## Attaching package: 'S4Vectors'## The following object is masked from 'package:utils':
##
## findMatches## The following objects are masked from 'package:base':
##
## I, expand.grid, unname## Loading required package: SummarizedExperiment## Loading required package: MatrixGenerics## Loading required package: matrixStats##
## Attaching package: 'MatrixGenerics'## The following objects are masked from 'package:matrixStats':
##
## colAlls, colAnyNAs, colAnys, colAvgsPerRowSet, colCollapse,
## colCounts, colCummaxs, colCummins, colCumprods, colCumsums,
## colDiffs, colIQRDiffs, colIQRs, colLogSumExps, colMadDiffs,
## colMads, colMaxs, colMeans2, colMedians, colMins, colOrderStats,
## colProds, colQuantiles, colRanges, colRanks, colSdDiffs, colSds,
## colSums2, colTabulates, colVarDiffs, colVars, colWeightedMads,
## colWeightedMeans, colWeightedMedians, colWeightedSds,
## colWeightedVars, rowAlls, rowAnyNAs, rowAnys, rowAvgsPerColSet,
## rowCollapse, rowCounts, rowCummaxs, rowCummins, rowCumprods,
## rowCumsums, rowDiffs, rowIQRDiffs, rowIQRs, rowLogSumExps,
## rowMadDiffs, rowMads, rowMaxs, rowMeans2, rowMedians, rowMins,
## rowOrderStats, rowProds, rowQuantiles, rowRanges, rowRanks,
## rowSdDiffs, rowSds, rowSums2, rowTabulates, rowVarDiffs, rowVars,
## rowWeightedMads, rowWeightedMeans, rowWeightedMedians,
## rowWeightedSds, rowWeightedVars## Loading required package: GenomicRanges## Loading required package: IRanges## Loading required package: GenomeInfoDb## Loading required package: Biobase## Welcome to Bioconductor
##
## Vignettes contain introductory material; view with
## 'browseVignettes()'. To cite Bioconductor, see
## 'citation("Biobase")', and for packages 'citation("pkgname")'.##
## Attaching package: 'Biobase'## The following object is masked from 'package:MatrixGenerics':
##
## rowMedians## The following objects are masked from 'package:matrixStats':
##
## anyMissing, rowMediansThe main input data is a [SummarizedExperiment] (https://bioconductor.org/packages/release/bioc/html/ SummarizedExperiment.html) object which contains an assay with count data. Briefly, a SummarizedExperiment container contains one or more assays, each represented by a matrix-like object of numeric or other mode. The rows typically represent genomic ranges of interest and the columns represent samples. For SeqGate, the SummarizedExperiment object must contain at least one assay of numeric counts, and a DataFrame describing the columns, in particular a column telling the biological condition the sample belongs to. To apply SeqGate, the SummarizedExperiment object, the assay name and the column describing the condition of each sample in the colData dataframe, must be given.
Let’s load some toy data set. This data set is an extract from a human transcriptome dataset produced by Strub et al. (2011), in which human cells expressing the Microphtalmia Transcription Factor (MiTF) are compared to cells in which the MiTF is repressed. The extract counts 1,000 genes with expression measured in 3 samples for each biological condition (the full table of read counts is available in the Supplementary materials of Dillies, M.A. et al. (2012)).
## A1 A2 B1 B2 A3 B3
## G000001 576 1048 993 756 825 1266
## G000002 0 0 0 0 0 0
## G000003 2590 3401 2674 2078 2879 3221
## G000004 109 35 255 169 111 184
## G000005 293 375 1289 719 415 1139
## G000006 0 0 8 0 0 4And now we define a vector indicating the biological condition corresponding to each column of data_MiTF_1000. Here the two biological conditions are ‘A’ and ‘B’.
The toy dataset that we have just loaded is not yet a SummarizedExperiment object, such as required in Seqgate input. We thus need to create it, from the count matrix and the biological condition annotation.
By default, SeqGate only needs the SummarizedExperiment object along with the name of the assay we want to work with, and the name of the column which contains the biological conditions annotation. Thus, we can apply the SeqGate method filtering, by calling the following code:
## assay(countsSE,counts) dataframe has
## been converted to a matrix.## The SeqGate filter parameters applied are:
## prop0=0.666666666666667
## percentile=0.9
## propUpThresh=0.9
## The applied threshold is 13.As a result, the input SummarizedExperiment object now includes a new column in the rowData DataFrame, named onFilter. This column is a logical vector that indicates if the gene should be kept after filtering (TRUE) or not (FALSE). The metadata of the object also include a new element, named “threshold”, which gives the value of the applied threshold.
Thus, to get the matrix of features intended to be kept for the downstream analysis:
## A1 A2 B1 B2 A3 B3
## G000001 576 1048 993 756 825 1266
## G000003 2590 3401 2674 2078 2879 3221
## G000004 109 35 255 169 111 184
## G000005 293 375 1289 719 415 1139
## G000008 1098 1281 1409 1284 1015 1640
## G000009 993 953 1488 1065 1293 1599## [1] 748 6To get the applied threshold:
## 90%
## 13We can also get the matrix of filtered genes:
## A1 A2 B1 B2 A3 B3
## G000002 0 0 0 0 0 0
## G000006 0 0 8 0 0 4
## G000007 51 10 56 0 38 46
## G000014 0 35 48 0 55 9
## G000016 13 0 0 0 0 0
## G000021 0 0 0 0 0 10To conclude, we can see that, from the initial set of 1,000 genes, 748 have been kept, after the application of a threshold of 13: all genes having less than 90% replicates with less than 13 reads are discarded.
Besides the three mandatory parameters described above, the applySeqGate function also have three other parameters, that can be set to refine the filtering:
By default, ‘prop0’ is set to the maximum number of replicates minus one, divided by the maximum number of replicates. In the example above, as we have 3 replicates in both conditions, the maximum number of replicates is 3. Thus, the parameter ‘prop0’ is set to 2/3. This means that we consider that the gene is lowly expressed if it has 2 zeros among its 3 replicates.
The distribution of maximum counts from all the lowly expressed genes (selected according to ‘prop0’) is then computed. The idea is to see how high a count can be in a replicate alongside a zero in another replicate. In order to introduce flexibility, we do not simply take the maximum count of the distribution but a ‘percentile’ of this distribution. By default, when the number of replicates in at least one condition is below 5, ‘percentile’ is set to 0.9. In the above example, the 90th percentile of the distribution of maximum counts seen alongside a zero is 13, and this is the threshold that we will apply in order to actually filter the lowly expressed genes.
Finally, the filter is applied according to a last parameter: propUpThresh. SeqGate does keep those genes whose counts are above the computed threshold in at least ‘propUpThresh’ replicates, in at least one condition. Still in the example used precedently, this means that all genes whose counts are above 13 in 3 x 0.9 = 2.7 replicates, are kept. As it is not possible to consider 2.7 replicates, the value is rounded to the next integer, that is 3 in this case. Finally in this example, a gene is kept if all its 3 replicates have a count above 13, in at least one condition.
Default value for all customizable parameters have been set through extensive simulation batch testing and can be considered as adequate in most situations. However, one may consider that the default parameters are not suited to its experiment. In that case, custom values can be given:
counts_strub <- applySeqGate(counts_strub,"counts","Conditions",
prop0=1/3,
percentile=0.8,
propUpThresh=0.5)## The SeqGate filter parameters applied are:
## prop0=0.333333333333333
## percentile=0.8
## propUpThresh=0.5
## The applied threshold is 18.This time, from the initial set of 1,000 genes, 765 have been kept, after the application of a threshold of 18.
## R version 4.4.1 (2024-06-14)
## Platform: x86_64-pc-linux-gnu
## Running under: Ubuntu 24.04.1 LTS
##
## Matrix products: default
## BLAS: /usr/lib/x86_64-linux-gnu/openblas-pthread/libblas.so.3
## LAPACK: /usr/lib/x86_64-linux-gnu/openblas-pthread/libopenblasp-r0.3.26.so; LAPACK version 3.12.0
##
## locale:
## [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
## [3] LC_TIME=en_US.UTF-8 LC_COLLATE=C
## [5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
## [7] LC_PAPER=en_US.UTF-8 LC_NAME=C
## [9] LC_ADDRESS=C LC_TELEPHONE=C
## [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C
##
## time zone: Etc/UTC
## tzcode source: system (glibc)
##
## attached base packages:
## [1] stats4 stats graphics grDevices utils datasets methods
## [8] base
##
## other attached packages:
## [1] SeqGate_1.17.0 SummarizedExperiment_1.36.0
## [3] Biobase_2.67.0 GenomicRanges_1.59.0
## [5] GenomeInfoDb_1.43.0 IRanges_2.41.0
## [7] MatrixGenerics_1.19.0 matrixStats_1.4.1
## [9] S4Vectors_0.44.0 BiocGenerics_0.53.0
## [11] knitr_1.48 BiocStyle_2.35.0
##
## loaded via a namespace (and not attached):
## [1] Matrix_1.7-1 jsonlite_1.8.9 compiler_4.4.1
## [4] BiocManager_1.30.25 crayon_1.5.3 jquerylib_0.1.4
## [7] yaml_2.3.10 fastmap_1.2.0 lattice_0.22-6
## [10] R6_2.5.1 XVector_0.46.0 S4Arrays_1.6.0
## [13] DelayedArray_0.33.1 maketools_1.3.1 GenomeInfoDbData_1.2.13
## [16] bslib_0.8.0 rlang_1.1.4 cachem_1.1.0
## [19] xfun_0.48 sass_0.4.9 sys_3.4.3
## [22] SparseArray_1.6.0 cli_3.6.3 zlibbioc_1.52.0
## [25] digest_0.6.37 grid_4.4.1 lifecycle_1.0.4
## [28] evaluate_1.0.1 buildtools_1.0.0 abind_1.4-8
## [31] rmarkdown_2.28 httr_1.4.7 tools_4.4.1
## [34] htmltools_0.5.8.1 UCSC.utils_1.2.0