Computing Dynamic SWATH Windows
Witold E. Wolski, Bernd Roschitzki
2021-12-07
Introduction
Function optimizes Extraction windows for DIA/SWATH so we have the same number of precursor per window. This optimization is based on spectral library data or non redundant .blib files (Bibliospec).
Prerequisites
Constant with method
data("masses")
cdsw <- Cdsw(masses , nbins = 25, digits = 1)
cdsw$plot()
knitr::kable(cdsw$asTable())| from | to | mid | width | counts |
|---|---|---|---|---|
| 349.63 | 384.62 | 367.125 | 34.99 | 6688 |
| 383.62 | 418.62 | 401.120 | 35.00 | 8357 |
| 417.62 | 452.61 | 435.115 | 34.99 | 9661 |
| 451.61 | 486.61 | 469.110 | 35.00 | 10452 |
| 485.61 | 520.60 | 503.105 | 34.99 | 10725 |
| 519.60 | 554.59 | 537.095 | 34.99 | 10837 |
| 553.59 | 588.59 | 571.090 | 35.00 | 10433 |
| 587.59 | 622.58 | 605.085 | 34.99 | 9750 |
| 621.58 | 656.58 | 639.080 | 35.00 | 9276 |
| 655.58 | 690.57 | 673.075 | 34.99 | 8406 |
| 689.57 | 724.56 | 707.065 | 34.99 | 7848 |
| 723.56 | 758.56 | 741.060 | 35.00 | 7116 |
| 757.56 | 792.55 | 775.055 | 34.99 | 6355 |
| 791.55 | 826.55 | 809.050 | 35.00 | 5666 |
| 825.55 | 860.54 | 843.045 | 34.99 | 4923 |
| 859.54 | 894.53 | 877.035 | 34.99 | 4359 |
| 893.53 | 928.53 | 911.030 | 35.00 | 3807 |
| 927.53 | 962.52 | 945.025 | 34.99 | 3344 |
| 961.52 | 996.52 | 979.020 | 35.00 | 2724 |
| 995.52 | 1030.51 | 1013.015 | 34.99 | 2357 |
| 1029.51 | 1064.50 | 1047.005 | 34.99 | 2042 |
| 1063.50 | 1098.50 | 1081.000 | 35.00 | 1807 |
| 1097.50 | 1132.49 | 1114.995 | 34.99 | 1313 |
| 1131.49 | 1166.49 | 1148.990 | 35.00 | 1088 |
| 1165.49 | 1200.48 | 1182.985 | 34.99 | 881 |
constError <- cdsw$error()Classical Method based on quantile
Same number of MS1 precursors in each window
cdsw$quantile_breaks()cdsw$plot()
knitr::kable(cdsw$asTable())| from | to | mid | width | counts | |
|---|---|---|---|---|---|
| 0% | 349.63 | 381.03 | 365.330 | 31.40 | 5956 |
| 4% | 380.03 | 406.71 | 393.370 | 26.68 | 6131 |
| 8% | 405.71 | 429.24 | 417.475 | 23.53 | 6070 |
| 12% | 428.24 | 450.05 | 439.145 | 21.81 | 6086 |
| 16% | 449.05 | 470.06 | 459.555 | 21.01 | 6095 |
| 20% | 469.06 | 488.80 | 478.930 | 19.74 | 6107 |
| 24% | 487.80 | 508.12 | 497.960 | 20.32 | 6173 |
| 28% | 507.12 | 526.81 | 516.965 | 19.69 | 6150 |
| 32% | 525.81 | 545.79 | 535.800 | 19.98 | 6166 |
| 36% | 544.79 | 565.29 | 555.040 | 20.50 | 6123 |
| 40% | 564.29 | 584.80 | 574.545 | 20.51 | 6139 |
| 44% | 583.80 | 605.12 | 594.460 | 21.32 | 6121 |
| 48% | 604.12 | 626.34 | 615.230 | 22.22 | 6113 |
| 52% | 625.34 | 648.36 | 636.850 | 23.02 | 6108 |
| 56% | 647.36 | 672.34 | 659.850 | 24.98 | 6074 |
| 60% | 671.34 | 696.53 | 683.935 | 25.19 | 6082 |
| 64% | 695.53 | 722.89 | 709.210 | 27.36 | 6054 |
| 68% | 721.89 | 751.40 | 736.645 | 29.51 | 6053 |
| 72% | 750.40 | 782.43 | 766.415 | 32.03 | 6023 |
| 76% | 781.43 | 817.40 | 799.415 | 35.97 | 5982 |
| 80% | 816.40 | 857.96 | 837.180 | 41.56 | 6026 |
| 84% | 856.96 | 905.62 | 881.290 | 48.66 | 5971 |
| 88% | 904.62 | 964.93 | 934.775 | 60.31 | 5943 |
| 92% | 963.93 | 1049.48 | 1006.705 | 85.55 | 5903 |
| 96% | 1048.48 | 1200.48 | 1124.480 | 152.00 | 5863 |
quantileError <- cdsw$error()Adjust windows
Using this method the window start and end is shifted to a mass range with as few MS1 peaks as possible.
knitr::kable(cdsw$optimizeWindows(maxbin = 10, plot = TRUE) )
| from | to | mid | width | counts |
|---|---|---|---|---|
| 350.13 | 380.95 | 365.54 | 30.82 | 5952 |
| 380.45 | 406.35 | 393.40 | 25.90 | 5932 |
| 406.05 | 429.05 | 417.55 | 23.00 | 5948 |
| 428.65 | 449.65 | 439.15 | 21.00 | 5891 |
| 449.45 | 469.65 | 459.55 | 20.20 | 5872 |
| 469.45 | 488.35 | 478.90 | 18.90 | 5860 |
| 488.15 | 508.05 | 498.10 | 19.90 | 6137 |
| 507.45 | 526.45 | 516.95 | 19.00 | 5892 |
| 526.15 | 545.45 | 535.80 | 19.30 | 6022 |
| 545.15 | 565.05 | 555.10 | 19.90 | 5992 |
| 564.55 | 584.45 | 574.50 | 19.90 | 5976 |
| 584.15 | 605.05 | 594.60 | 20.90 | 6035 |
| 604.55 | 626.05 | 615.30 | 21.50 | 5893 |
| 625.55 | 648.15 | 636.85 | 22.60 | 5987 |
| 647.55 | 672.15 | 659.85 | 24.60 | 6023 |
| 671.75 | 696.15 | 683.95 | 24.40 | 5890 |
| 695.55 | 722.55 | 709.05 | 27.00 | 5981 |
| 722.25 | 751.15 | 736.70 | 28.90 | 5932 |
| 750.65 | 782.15 | 766.40 | 31.50 | 5927 |
| 781.65 | 817.15 | 799.40 | 35.50 | 5944 |
| 816.65 | 857.55 | 837.10 | 40.90 | 5901 |
| 857.35 | 905.25 | 881.30 | 47.90 | 5904 |
| 905.05 | 964.65 | 934.85 | 59.60 | 5881 |
| 964.35 | 1049.15 | 1006.75 | 84.80 | 5864 |
| 1048.95 | 1200.05 | 1124.50 | 151.10 | 5843 |
Dynamic Swath Windows with Constraints.
- Mass range can be specified (mass_range)
- Maximal window size can be specified (max_window_size). This is because windows should not be to large because of optimal collision energy (personal communication by Bernd R.).
- Minimal window size can be specified (min_window_size).
- target number of windows can be specified (nr_windows).
- boundaries between windows are placed in regions were no precursors are observed.
cdsw$sampling_breaks(maxwindow = 100,plot = TRUE)
cdsw$plot()
knitr::kable(cdsw$asTable())| from | to | mid | width | counts | |
|---|---|---|---|---|---|
| 0% | 349.63 | 381.71 | 365.670 | 32.08 | 6113 |
| 4% | 380.71 | 408.41 | 394.560 | 27.70 | 6371 |
| 8% | 407.41 | 432.26 | 419.835 | 24.85 | 6578 |
| 12% | 431.26 | 454.93 | 443.095 | 23.67 | 6622 |
| 16% | 453.93 | 476.38 | 465.155 | 22.45 | 6625 |
| 20% | 475.38 | 497.11 | 486.245 | 21.73 | 6678 |
| 24% | 496.11 | 518.13 | 507.120 | 22.02 | 6763 |
| 28% | 517.13 | 538.79 | 527.960 | 21.66 | 6730 |
| 32% | 537.79 | 560.07 | 548.930 | 22.28 | 6771 |
| 36% | 559.07 | 581.30 | 570.185 | 22.23 | 6691 |
| 40% | 580.30 | 603.26 | 591.780 | 22.96 | 6621 |
| 44% | 602.26 | 626.18 | 614.220 | 23.92 | 6603 |
| 48% | 625.18 | 649.87 | 637.525 | 24.69 | 6572 |
| 52% | 648.87 | 675.24 | 662.055 | 26.37 | 6430 |
| 56% | 674.24 | 701.24 | 687.740 | 27.00 | 6402 |
| 60% | 700.24 | 728.90 | 714.570 | 28.66 | 6322 |
| 64% | 727.90 | 758.92 | 743.410 | 31.02 | 6244 |
| 68% | 757.92 | 791.39 | 774.655 | 33.47 | 6093 |
| 72% | 790.39 | 826.91 | 808.650 | 36.52 | 5891 |
| 76% | 825.91 | 866.98 | 846.445 | 41.07 | 5730 |
| 80% | 865.98 | 912.46 | 889.220 | 46.48 | 5469 |
| 84% | 911.46 | 963.54 | 937.500 | 52.08 | 5119 |
| 88% | 962.54 | 1026.85 | 994.695 | 64.31 | 4672 |
| 92% | 1025.85 | 1101.01 | 1063.430 | 75.16 | 4134 |
| 96% | 1100.01 | 1200.48 | 1150.245 | 100.47 | 3118 |
knitr::kable(cdsw$optimizeWindows(maxbin = 10, plot = TRUE) )
| from | to | mid | width | counts |
|---|---|---|---|---|
| 350.13 | 381.35 | 365.74 | 31.22 | 6053 |
| 381.05 | 408.05 | 394.55 | 27.00 | 6191 |
| 407.65 | 432.05 | 419.85 | 24.40 | 6448 |
| 431.65 | 454.45 | 443.05 | 22.80 | 6448 |
| 454.35 | 476.05 | 465.20 | 21.70 | 6389 |
| 475.85 | 496.85 | 486.35 | 21.00 | 6526 |
| 496.45 | 517.85 | 507.15 | 21.40 | 6587 |
| 517.45 | 538.45 | 527.95 | 21.00 | 6492 |
| 538.15 | 560.05 | 549.10 | 21.90 | 6688 |
| 559.55 | 581.05 | 570.30 | 21.50 | 6482 |
| 580.55 | 603.05 | 591.80 | 22.50 | 6526 |
| 602.55 | 626.05 | 614.30 | 23.50 | 6452 |
| 625.55 | 649.45 | 637.50 | 23.90 | 6335 |
| 649.25 | 675.15 | 662.20 | 25.90 | 6392 |
| 674.55 | 701.15 | 687.85 | 26.60 | 6293 |
| 700.55 | 728.55 | 714.55 | 28.00 | 6160 |
| 728.25 | 758.55 | 743.40 | 30.30 | 6119 |
| 758.25 | 791.25 | 774.75 | 33.00 | 6038 |
| 790.65 | 826.55 | 808.60 | 35.90 | 5815 |
| 826.25 | 866.55 | 846.40 | 40.30 | 5622 |
| 866.35 | 912.05 | 889.20 | 45.70 | 5408 |
| 911.85 | 963.15 | 937.50 | 51.30 | 5055 |
| 962.75 | 1026.35 | 994.55 | 63.60 | 4631 |
| 1026.25 | 1100.65 | 1063.45 | 74.40 | 4105 |
| 1100.45 | 1200.05 | 1150.25 | 99.60 | 3091 |
mixedError <- cdsw$error()Benchmarking of the methods.
We compare the optimal number of MS1 peaks per SWATH window (same in each window) with the numbers obtained by using all of the 3 methods implemented.
barplot(c(const = constError$score1, quantile = quantileError$score1, mixed = mixedError$score1),ylab = "Manhattan distance")
barplot(c(const = constError$score2, quantile = quantileError$score2, mixed = mixedError$score2),ylab = "Euclidean distance")
We can see that Method 3 has a relatively small error although it is able to fulfill constraints such as maximum window size.
Session info
## R version 4.1.1 (2021-08-10)
## Platform: x86_64-w64-mingw32/x64 (64-bit)
## Running under: Windows 10 x64 (build 19044)
##
## Matrix products: default
##
## locale:
## [1] LC_COLLATE=C
## [2] LC_CTYPE=English_United States.1252
## [3] LC_MONETARY=English_United States.1252
## [4] 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] dplyr_1.0.7 Matrix_1.3-4 prozor_0.3.1
##
## loaded via a namespace (and not attached):
## [1] Rcpp_1.0.7 highr_0.9 bslib_0.3.1
## [4] compiler_4.1.1 pillar_1.6.4 jquerylib_0.1.4
## [7] tools_4.1.1 bit_4.0.4 digest_0.6.28
## [10] docopt_0.7.1 jsonlite_1.7.2 evaluate_0.14
## [13] lifecycle_1.0.1 tibble_3.1.4 lattice_0.20-44
## [16] AhoCorasickTrie_0.1.2 pkgconfig_2.0.3 rlang_0.4.11
## [19] DBI_1.1.1 cli_3.1.0 parallel_4.1.1
## [22] yaml_2.2.1 xfun_0.26 fastmap_1.1.0
## [25] stringr_1.4.0 knitr_1.36 generics_0.1.1
## [28] sass_0.4.0 vctrs_0.3.8 hms_1.1.1
## [31] tidyselect_1.1.1 bit64_4.0.5 ade4_1.7-18
## [34] grid_4.1.1 glue_1.4.2 R6_2.5.1
## [37] fansi_0.5.0 vroom_1.5.6 rmarkdown_2.11
## [40] tzdb_0.1.2 purrr_0.3.4 readr_2.0.1
## [43] seqinr_4.2-8 magrittr_2.0.1 htmltools_0.5.2
## [46] ellipsis_0.3.2 MASS_7.3-54 assertthat_0.2.1
## [49] utf8_1.2.2 stringi_1.7.4 crayon_1.4.2