AAC and AUC values are the same

[islic]

Hello , I had to update my R in order to intersect two Psets. I also updated to PharmacoGx package to version 2.2.4. I had downloaded from the previous version auc_published and auc_recomputed values for GDSC and CTRPv2. When I updated the package auc was not availbale as a sensitivity measure but instead aac_recomputed was availbale for GDSC and aac_recomputed or aac_published for CTRPv2. However when I checked between the two versions auc_recomputed and aac_recomputed for GDSC were the same . I checked for CTRPv2 and it is the same case (aac_recomputed and auc_recomputed are the same and aac_published and auc_published are the same). By looking at the drugDoseResponseCurves these values are probably an indicator of the area above the curve. I would like to know which sensitivity measure do these values correspond to? (AUC or AAC) Below is my session info as well:

sessionInfo() R version 4.0.5 (2021-03-31) Platform: x86_64-w64-mingw32/x64 (64-bit) Running under: Windows 10 x64 (build 17763)

Matrix products: default

locale: [1] LC_COLLATE=English_United States.1252 LC_CTYPE=English_United States.1252
[3] LC_MONETARY=English_United States.1252 LC_NUMERIC=C
[5] LC_TIME=English_United States.1252

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

other attached packages: [1] PharmacoGx_2.2.4 CoreGx_1.2.0

loaded via a namespace (and not attached): [1] lsa_0.73.2 bitops_1.0-6 matrixStats_0.58.0
[4] RColorBrewer_1.1-2 GenomeInfoDb_1.26.7 SnowballC_0.7.0
[7] tools_4.0.5 utf8_1.2.1 R6_2.5.0
[10] DT_0.18 KernSmooth_2.23-18 sm_2.2-5.6
[13] DBI_1.1.1 BiocGenerics_0.36.1 colorspace_2.0-0
[16] tidyselect_1.1.1 gridExtra_2.3 curl_4.3
[19] compiler_4.0.5 Biobase_2.50.0 shinyjs_2.0.0
[22] DelayedArray_0.16.3 slam_0.1-48 caTools_1.18.2
[25] scales_1.1.1 relations_0.6-9 stringr_1.4.0
[28] digest_0.6.27 XVector_0.30.0 pkgconfig_2.0.3
[31] htmltools_0.5.1.1 plotrix_3.8-1 MatrixGenerics_1.2.1
[34] fastmap_1.1.0 limma_3.46.0 maps_3.3.0
[37] htmlwidgets_1.5.3 rlang_0.4.10 shiny_1.6.0
[40] visNetwork_2.0.9 generics_0.1.0 jsonlite_1.7.2
[43] BiocParallel_1.24.1 gtools_3.8.2 dplyr_1.0.5
[46] RCurl_1.98-1.3 magrittr_2.0.1 GenomeInfoDbData_1.2.4
[49] Matrix_1.3-2 Rcpp_1.0.6 celestial_1.4.6
[52] munsell_0.5.0 S4Vectors_0.28.1 fansi_0.4.2
[55] lifecycle_1.0.0 stringi_1.5.3 piano_2.6.0
[58] MASS_7.3-53.1 SummarizedExperiment_1.20.0 zlibbioc_1.36.0
[61] plyr_1.8.6 gplots_3.1.1 grid_4.0.5
[64] parallel_4.0.5 promises_1.2.0.1 shinydashboard_0.7.1
[67] crayon_1.4.1 lattice_0.20-41 mapproj_1.2.7
[70] pillar_1.6.0 fgsea_1.16.0 tcltk_4.0.5
[73] igraph_1.2.6 GenomicRanges_1.42.0 reshape2_1.4.4
[76] marray_1.68.0 stats4_4.0.5 fastmatch_1.1-0
[79] NISTunits_1.0.1 glue_1.4.2 downloader_0.4
[82] data.table_1.14.0 vctrs_0.3.7 httpuv_1.5.5
[85] gtable_0.3.0 RANN_2.6.1 purrr_0.3.4
[88] assertthat_0.2.1 ggplot2_3.3.3 mime_0.10
[91] xtable_1.8-4 pracma_2.3.3 later_1.1.0.1
[94] tibble_3.1.0 IRanges_2.24.1 sets_1.0-18
[97] cluster_2.1.1 ellipsis_0.3.1 magicaxis_2.2.1

[ChristopherEeles]

Hi @islic,

We version control the creation of all of our PharmacoSet objects using ORCESTRA. Could you provide the return of dateCreated(PSet) for the two data objects you are using? We can use this date to look up the documentation for the specific objects you are working with.

I suspect that the values in the auc_* columns are actually AAC, as that is the standard metric we use to assess dose-response curves. We made that change a while ago, so it is likely we forgot to update the column names at that time and the change you are seeing just corrects the column names to reflect what they actually contain. If you provide the dates I can double-check that my guess is correct.

Best, Christopher Eeles Software Developer

[islic]

Hello @ChristopherEeles , Below are the dates of the two Psets that contain AUC values which I was using previously dateCreated(GDSC) [1] "Thu Aug 18 16:10:06 2016"

dateCreated(CTRPv2) [1] "Wed May 11 10:20:02 2016" Here are the dates of the same Psets but after updating the package that contain AAC values only : dateCreated(GDSC1) [1] "Wed Jun 24 02:01:01 2020" dateCreated(CTRP1) [1] "Tue Jun 23 23:32:05 2020"

All the Best, Isli

[p-smirnov]

Hi Isli,

You are correct in that the 2016 auc values were actually AAC values for the viability curve.

The auc/AAC confusion arises due to the fact that you can look at dose-viability or dose-inhibition curves. A couple years ago we standardized (following the large CTRP and GDSC projects) to refer to dose-viability curves, and therefore names were changed to area above the curve.

This also means that higher numbers = more sensitive, which makes communicating results simpler.

Best, Petr

[islic]

Hello, Thank you for the reply. It makes sense that the auc_recomputed values from the 2016 version are actually aac values. However I noticed that for CTRP the 2016 auc_published and the 2020 aac_published values are the same as well and the range of these values is not from 0 -1 but from 0-16. Should I regard these values as aac or auc values and do you have any idea to why their range isn`t from 0 to 1 ? In the picture below the red line is the 2016 CTRP values and the blue line is the 2020 CTRP values and as you can see the numbers are the same.

Best , Isli

[p-smirnov]

@islic
Looks like that is a naming mistake, they should be auc_published. Will fix it asap! The published values were not normalized by concentration range by the original study authors, and we kept them as such. Therefore, for each experiment the maximum possible value is different.

[ChristopherEeles]

@p-smirnov,

Was this naming issue ever fixed?