Read SCP sensitive to the order the quantCol rows are provided in the sample annotation file

[edemmott]

I've sent Laurent an email with some more details but in brief, readSCP is sensitive to row order in the sample annotation file. Two files with identical data, but with the rows reordered based on the quantCol give different results:

Example 1, ordered as below gives poor results.
Reporter.intensity.1
Reporter.intensity.10
Reporter.intensity.11
Reporter.intensity.12
Reporter.intensity.13
Reporter.intensity.14
Reporter.intensity.15
Reporter.intensity.16
Reporter.intensity.17
Reporter.intensity.18
Reporter.intensity.2
Reporter.intensity.3
Reporter.intensity.4
Reporter.intensity.5
Reporter.intensity.6
Reporter.intensity.7
Reporter.intensity.8
Reporter.intensity.9

Example 2, ordered below, gives good results.
Reporter.intensity.1
Reporter.intensity.2
Reporter.intensity.3
Reporter.intensity.4
Reporter.intensity.5
Reporter.intensity.6
Reporter.intensity.7
Reporter.intensity.8
Reporter.intensity.9
Reporter.intensity.10
Reporter.intensity.11
Reporter.intensity.12
Reporter.intensity.13
Reporter.intensity.14
Reporter.intensity.15
Reporter.intensity.16
Reporter.intensity.17
Reporter.intensity.18

[edemmott]

It looks like with example 1, the quantCols accessible with assay() are misassigned

[cvanderaa]

Hi @edemmott,

Thanks for reporting the bug!

I don't know if something has already been discussed by mail with Laurent. Are you using the latest stable version of scp, ie scp >= 1.16.0?

If so, I would like to solve this bug as quickly as possible (actually in QFeatures). Would you have a small reproducible example, or at least the code you used to run your examples (I can mock data) ?

[edemmott]

Will drop you/laurent an email with a link to the data/some of our .rmd and the documents.

Yes using 1.16. SessionInfo below:

R version 4.4.1 (2024-06-14)
Platform: aarch64-apple-darwin20
Running under: macOS Sonoma 14.5

Matrix products: default
BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

time zone: Europe/London
tzcode source: internal

attached base packages:
[1] stats4 stats graphics grDevices utils datasets methods base

other attached packages:
[1] iSEEu_1.18.0 iSEEhex_1.8.0 imager_1.0.2 magrittr_2.0.3
[5] png_0.1-8 shiny_1.9.1 iSEE_2.18.0 bluster_1.16.0
[9] EnsDb.Hsapiens.v86_2.99.0 ensembldb_2.30.0 AnnotationFilter_1.30.0 GenomicFeatures_1.58.0
[13] AnnotationDbi_1.68.0 SCP.replication_0.2.1 data.table_1.16.2 scater_1.34.0
[17] scuttle_1.16.0 sva_3.54.0 BiocParallel_1.39.0 genefilter_1.88.0
[21] mgcv_1.9-1 nlme_3.1-166 patchwork_1.3.0 lubridate_1.9.3
[25] forcats_1.0.0 stringr_1.5.1 dplyr_1.1.4 purrr_1.0.2
[29] readr_2.1.5 tidyr_1.3.1 tibble_3.2.1 ggplot2_3.5.1
[33] tidyverse_2.0.0 limma_3.62.1 scpdata_1.13.0 ExperimentHub_2.14.0
[37] AnnotationHub_3.14.0 BiocFileCache_2.14.0 dbplyr_2.5.0 SingleCellExperiment_1.28.1
[41] scp_1.16.0 QFeatures_1.16.0 MultiAssayExperiment_1.32.0 SummarizedExperiment_1.36.0
[45] Biobase_2.66.0 GenomicRanges_1.58.0 GenomeInfoDb_1.42.0 IRanges_2.40.0
[49] S4Vectors_0.44.0 BiocGenerics_0.52.0 MatrixGenerics_1.18.0 matrixStats_1.4.1
[53] Seurat_5.1.0 SeuratObject_5.0.2 sp_2.1-4 QuantQC_0.1.0

[edemmott]

Email sent to you both with examples and dataset.

[cvanderaa]

I'm making progress, and I confirm that what you are seeing is a bug. Here is a minimal reproducible example:

library(scp)
ad <- matrix(1:75, ncol = 5, dimnames = list(NULL, rev(paste0("quantCol", 1:5))))
ad <- as.data.frame(ad)
ad$runCol <- rep(paste0("run", 1:3), each = 5)
cd <- data.frame(quantCols = rep(paste0("quantCol", 1:5), 3),
                 runCol = rep(paste0("run", 1:3), each = 5))
scp <- readSCP(assayData = ad,
               colData = cd,
               runCol = "runCol")

ad[ad$runCol == ad$runCol[[1]], ]

  quantCol5 quantCol4 quantCol3 quantCol2 quantCol1 runCol
1         1        16        31        46        61   run1
2         2        17        32        47        62   run1
3         3        18        33        48        63   run1
4         4        19        34        49        64   run1
5         5        20        35        50        65   run1

assay(scp, 1)

 run1_quantCol1 run1_quantCol2 run1_quantCol3 run1_quantCol4 run1_quantCol5
1              1             16             31             46             61
2              2             17             32             47             62
3              3             18             33             48             63
4              4             19             34             49             64
5              5             20             35             50             65

The second table has wrong column names and should match the first table.

This is indeed a QFeatures issue. After running the code in debug mode, here is the problematic line.

I'll try to work on it asap.

[cvanderaa]

We are making progress, a preliminary fix is available here: https://github.com/cvanderaa/QFeatures. We still need some time to integrate this in QFeatures' official devel release.