diff --git a/NAMESPACE b/NAMESPACE index a4708cc3..15abcb3d 100755 --- a/NAMESPACE +++ b/NAMESPACE @@ -217,6 +217,7 @@ importFrom(dplyr,bind_cols) importFrom(dplyr,bind_rows) importFrom(dplyr,case_when) importFrom(dplyr,desc) +importFrom(dplyr,distinct) importFrom(dplyr,everything) importFrom(dplyr,filter) importFrom(dplyr,filter_all) diff --git a/R/freq.R b/R/freq.R index 1f8f5aac..d498932b 100755 --- a/R/freq.R +++ b/R/freq.R @@ -682,7 +682,7 @@ format_header <- function(x, markdown = FALSE, decimal.mark = ".", big.mark = ", # numeric values if (has_length == TRUE & any(x_class %in% c("double", "integer", "numeric", "raw", "single"))) { header$sd <- paste0(header$sd, " (CV: ", header$cv, ", MAD: ", header$mad, ")") - header$fivenum <- paste0(paste(header$fivenum, collapse = " | "), " (IQR: ", header$IQR, ", CQV: ", header$cqv, ")") + header$fivenum <- paste0(paste(trimws(header$fivenum), collapse = " | "), " (IQR: ", header$IQR, ", CQV: ", header$cqv, ")") header$outliers_total <- paste0(header$outliers_total, " (unique count: ", header$outliers_unique, ")") header <- header[!names(header) %in% c("cv", "mad", "IQR", "cqv", "outliers_unique")] } diff --git a/R/mo.R b/R/mo.R index df08ec54..66deefa2 100755 --- a/R/mo.R +++ b/R/mo.R @@ -165,30 +165,44 @@ #' mutate(mo = as.mo(paste(genus, species))) #' } as.mo <- function(x, Becker = FALSE, Lancefield = FALSE, allow_uncertain = TRUE, reference_df = get_mo_source()) { + if (!"AMR" %in% base::.packages()) { + library("AMR") + # check onLoad() in R/zzz.R: data tables are created there. + } + if (all(x %in% AMR::microorganisms$mo) & isFALSE(Becker) & isFALSE(Lancefield) & is.null(reference_df)) { y <- x - } else if (all(x %in% AMR::microorganisms$fullname) - & isFALSE(Becker) - & isFALSE(Lancefield) - & is.null(reference_df)) { - # we need special treatment for very prevalent full names, they are likely! + + } else if (all(tolower(x) %in% microorganismsDT$fullname_lower) + & isFALSE(Becker) + & isFALSE(Lancefield) + & is.null(reference_df)) { + # we need special treatment for very prevalent full names, they are likely! (case insensitive) # e.g. as.mo("Staphylococcus aureus") - y <- microorganismsDT[prevalence == 1][data.table(fullname = x), on = "fullname", "mo"][[1]] + y <- microorganismsDT[prevalence == 1][data.table(fullname_lower = tolower(x)), + on = "fullname_lower", + "mo"][[1]] if (any(is.na(y))) { - y[is.na(y)] <- microorganismsDT[prevalence == 2][data.table(fullname = x[is.na(y)]), on = "fullname", "mo"][[1]] + y[is.na(y)] <- microorganismsDT[prevalence == 2][data.table(fullname_lower = tolower(x[is.na(y)])), + on = "fullname_lower", + "mo"][[1]] } if (any(is.na(y))) { - y[is.na(y)] <- microorganismsDT[prevalence == 3][data.table(fullname = x[is.na(y)]), on = "fullname", "mo"][[1]] + y[is.na(y)] <- microorganismsDT[prevalence == 3][data.table(fullname_lower = tolower(x[is.na(y)])), + on = "fullname_lower", + "mo"][[1]] } + } else { # will be checked for mo class in validation and uses exec_as.mo internally if necessary y <- mo_validate(x = x, property = "mo", Becker = Becker, Lancefield = Lancefield, allow_uncertain = allow_uncertain, reference_df = reference_df) } + structure(.Data = y, class = "mo") } @@ -198,7 +212,7 @@ is.mo <- function(x) { identical(class(x), "mo") } -#' @importFrom dplyr %>% pull left_join n_distinct progress_estimated filter +#' @importFrom dplyr %>% pull left_join n_distinct progress_estimated filter distinct #' @importFrom data.table data.table as.data.table setkey #' @importFrom crayon magenta red blue silver italic has_color exec_as.mo <- function(x, Becker = FALSE, Lancefield = FALSE, @@ -298,22 +312,30 @@ exec_as.mo <- function(x, Becker = FALSE, Lancefield = FALSE, # existing mo codes when not looking for property "mo", like mo_genus("B_ESCHR_COL") y <- microorganismsDT[prevalence == 1][data.table(mo = x), on = "mo", ..property][[1]] if (any(is.na(y))) { - y[is.na(y)] <- microorganismsDT[prevalence == 2][data.table(mo = x[is.na(y)]), on = "mo", ..property][[1]] + y[is.na(y)] <- microorganismsDT[prevalence == 2][data.table(mo = x[is.na(y)]), + on = "mo", + ..property][[1]] } if (any(is.na(y))) { - y[is.na(y)] <- microorganismsDT[prevalence == 3][data.table(mo = x[is.na(y)]), on = "mo", ..property][[1]] + y[is.na(y)] <- microorganismsDT[prevalence == 3][data.table(mo = x[is.na(y)]), + on = "mo", + ..property][[1]] } x <- y - } else if (all(x %in% AMR::microorganisms$fullname)) { + } else if (all(tolower(x) %in% microorganismsDT$fullname_lower)) { # we need special treatment for very prevalent full names, they are likely! # e.g. as.mo("Staphylococcus aureus") - y <- microorganismsDT[prevalence == 1][data.table(fullname = x), on = "fullname", ..property][[1]] + y <- microorganismsDT[prevalence == 1][data.table(fullname_lower = tolower(x)), on = "fullname_lower", ..property][[1]] if (any(is.na(y))) { - y[is.na(y)] <- microorganismsDT[prevalence == 2][data.table(fullname = x[is.na(y)]), on = "fullname", ..property][[1]] + y[is.na(y)] <- microorganismsDT[prevalence == 2][data.table(fullname_lower = tolower(x[is.na(y)])), + on = "fullname_lower", + ..property][[1]] } if (any(is.na(y))) { - y[is.na(y)] <- microorganismsDT[prevalence == 3][data.table(fullname = x[is.na(y)]), on = "fullname", ..property][[1]] + y[is.na(y)] <- microorganismsDT[prevalence == 3][data.table(fullname_lower = tolower(x[is.na(y)])), + on = "fullname_lower", + ..property][[1]] } x <- y @@ -521,13 +543,13 @@ exec_as.mo <- function(x, Becker = FALSE, Lancefield = FALSE, # FIRST TRY FULLNAMES AND CODES # if only genus is available, return only genus if (all(!c(x[i], x_trimmed[i]) %like% " ")) { - found <- microorganismsDT[tolower(fullname) %in% tolower(c(x_species[i], x_trimmed_species[i])), ..property][[1]] + found <- microorganismsDT[fullname_lower %in% tolower(c(x_species[i], x_trimmed_species[i])), ..property][[1]] if (length(found) > 0) { x[i] <- found[1L] next } if (nchar(x_trimmed[i]) >= 6) { - found <- microorganismsDT[tolower(fullname) %like% paste0(x_withspaces_start_only[i], "[a-z]+ species"), ..property][[1]] + found <- microorganismsDT[fullname_lower %like% paste0(x_withspaces_start_only[i], "[a-z]+ species"), ..property][[1]] if (length(found) > 0) { x[i] <- found[1L] next @@ -564,13 +586,13 @@ exec_as.mo <- function(x, Becker = FALSE, Lancefield = FALSE, e.x_withspaces_start_only, f.x_withspaces_end_only) { - found <- data_to_check[tolower(fullname) %in% tolower(c(a.x_backup, b.x_trimmed)), ..property][[1]] + found <- data_to_check[fullname_lower %in% tolower(c(a.x_backup, b.x_trimmed)), ..property][[1]] # most probable: is exact match in fullname if (length(found) > 0) { return(found[1L]) } - found <- data_to_check[tolower(fullname) == tolower(c.x_trimmed_without_group), ..property][[1]] + found <- data_to_check[fullname_lower == tolower(c.x_trimmed_without_group), ..property][[1]] if (length(found) > 0) { return(found[1L]) } @@ -664,7 +686,7 @@ exec_as.mo <- function(x, Becker = FALSE, Lancefield = FALSE, # MISCELLANEOUS ---- # look for old taxonomic names ---- - found <- microorganisms.oldDT[tolower(fullname) == tolower(x_backup[i]) + found <- microorganisms.oldDT[fullname_lower == tolower(x_backup[i]) | fullname %like% x_withspaces_start_end[i],] if (NROW(found) > 0) { col_id_new <- found[1, col_id_new] @@ -693,7 +715,7 @@ exec_as.mo <- function(x, Becker = FALSE, Lancefield = FALSE, if (nchar(b.x_trimmed) > 4 & !b.x_trimmed %like% " ") { if (!grepl("^[A-Z][a-z]+", b.x_trimmed, ignore.case = FALSE)) { # not when input is like Genustext, because then Neospora would lead to Actinokineospora - found <- microorganismsDT[tolower(fullname) %like% paste(b.x_trimmed, "species"), ..property][[1]] + found <- microorganismsDT[fullname_lower %like% paste(b.x_trimmed, "species"), ..property][[1]] if (length(found) > 0) { x[i] <- found[1L] uncertainties <<- rbind(uncertainties, diff --git a/R/zzz.R b/R/zzz.R index 5403922f..1bd04758 100755 --- a/R/zzz.R +++ b/R/zzz.R @@ -28,6 +28,7 @@ if (!all(c("microorganismsDT", "microorganisms.oldDT") %in% ls(envir = asNamespace("AMR")))) { microorganisms.oldDT <- as.data.table(AMR::microorganisms.old) + microorganisms.oldDT$fullname_lower <- tolower(microorganisms.oldDT$fullname) setkey(microorganisms.oldDT, col_id, fullname) assign(x = "microorganisms", @@ -84,6 +85,7 @@ make <- function() { #' @importFrom data.table as.data.table setkey make_DT <- function() { microorganismsDT <- as.data.table(make()) + microorganismsDT$fullname_lower <- tolower(microorganismsDT$fullname) setkey(microorganismsDT, kingdom, prevalence, diff --git a/docs/articles/AMR.html b/docs/articles/AMR.html index 915b8c7d..3e234c50 100644 --- a/docs/articles/AMR.html +++ b/docs/articles/AMR.html @@ -327,68 +327,68 @@
So, we can draw at least two conclusions immediately. From a data scientist perspective, the data looks clean: only values M
and F
. From a researcher perspective: there are slightly more men. Nothing we didn’t already know.
The data is already quite clean, but we still need to transform some variables. The bacteria
column now consists of text, and we want to add more variables based on microbial IDs later on. So, we will transform this column to valid IDs. The mutate()
function of the dplyr
package makes this really easy:
data <- data %>%
@@ -443,10 +443,10 @@
#> Kingella kingae (no changes)
#>
#> EUCAST Expert Rules, Intrinsic Resistance and Exceptional Phenotypes (v3.1, 2016)
-#> Table 1: Intrinsic resistance in Enterobacteriaceae (1291 changes)
+#> Table 1: Intrinsic resistance in Enterobacteriaceae (1256 changes)
#> Table 2: Intrinsic resistance in non-fermentative Gram-negative bacteria (no changes)
#> Table 3: Intrinsic resistance in other Gram-negative bacteria (no changes)
-#> Table 4: Intrinsic resistance in Gram-positive bacteria (2787 changes)
+#> Table 4: Intrinsic resistance in Gram-positive bacteria (2821 changes)
#> Table 8: Interpretive rules for B-lactam agents and Gram-positive cocci (no changes)
#> Table 9: Interpretive rules for B-lactam agents and Gram-negative rods (no changes)
#> Table 10: Interpretive rules for B-lactam agents and other Gram-negative bacteria (no changes)
@@ -462,9 +462,9 @@
#> Non-EUCAST: piperacillin/tazobactam = S where piperacillin = S (no changes)
#> Non-EUCAST: trimethoprim/sulfa = S where trimethoprim = S (no changes)
#>
-#> => EUCAST rules affected 7,442 out of 20,000 rows
+#> => EUCAST rules affected 7,457 out of 20,000 rows
#> -> added 0 test results
-#> -> changed 4,078 test results (0 to S; 0 to I; 4,078 to R)
So only 28.5% is suitable for resistance analysis! We can now filter on it with the filter()
function, also from the dplyr
package:
Only 1 isolates are marked as ‘first’ according to CLSI guideline. But when reviewing the antibiogram, it is obvious that some isolates are absolutely different strains and should be included too. This is why we weigh isolates, based on their antibiogram. The key_antibiotics()
function adds a vector with 18 key antibiotics: 6 broad spectrum ones, 6 small spectrum for Gram negatives and 6 small spectrum for Gram positives. These can be defined by the user.
Only 2 isolates are marked as ‘first’ according to CLSI guideline. But when reviewing the antibiogram, it is obvious that some isolates are absolutely different strains and should be included too. This is why we weigh isolates, based on their antibiogram. The key_antibiotics()
function adds a vector with 18 key antibiotics: 6 broad spectrum ones, 6 small spectrum for Gram negatives and 6 small spectrum for Gram positives. These can be defined by the user.
If a column exists with a name like ‘key(…)ab’ the first_isolate()
function will automatically use it and determine the first weighted isolates. Mind the NOTEs in below output:
data <- data %>%
mutate(keyab = key_antibiotics(.)) %>%
@@ -637,7 +637,7 @@
#> NOTE: Using column `patient_id` as input for `col_patient_id`.
#> NOTE: Using column `keyab` as input for `col_keyantibiotics`. Use col_keyantibiotics = FALSE to prevent this.
#> [Criterion] Inclusion based on key antibiotics, ignoring I.
-#> => Found 15,861 first weighted isolates (79.3% of total)
isolate | @@ -654,104 +654,104 @@|||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1 | -2010-02-12 | -I9 | +2010-01-18 | +I8 | B_ESCHR_COL | +I | S | S | S | -R | TRUE | TRUE | |
2 | -2010-02-12 | -I9 | +2010-03-30 | +I8 | B_ESCHR_COL | -R | S | +I | S | S | FALSE | -TRUE | +FALSE |
3 | -2010-02-22 | -I9 | +2010-10-05 | +I8 | B_ESCHR_COL | R | -S | R | S | +S | FALSE | TRUE | |
4 | -2010-03-05 | -I9 | +2011-01-26 | +I8 | B_ESCHR_COL | S | -S | +R | R | S | -FALSE | +TRUE | TRUE |
5 | -2010-03-08 | -I9 | +2011-02-01 | +I8 | B_ESCHR_COL | S | S | -R | -R | +S | +S | FALSE | TRUE |
6 | -2010-03-17 | -I9 | +2011-03-08 | +I8 | B_ESCHR_COL | +R | S | -S | -S | +R | S | FALSE | TRUE |
7 | -2010-05-03 | -I9 | +2011-04-11 | +I8 | B_ESCHR_COL | S | -S | +R | S | S | FALSE | -FALSE | +TRUE |
8 | -2010-07-03 | -I9 | +2011-04-23 | +I8 | B_ESCHR_COL | S | S | -S | +R | S | FALSE | -FALSE | +TRUE |
9 | -2010-09-11 | -I9 | +2011-06-21 | +I8 | B_ESCHR_COL | R | S | @@ -762,11 +762,11 @@||||||
10 | -2010-09-24 | -I9 | +2011-07-06 | +I8 | B_ESCHR_COL | S | -R | +S | S | S | FALSE | @@ -774,11 +774,11 @@
Instead of 1, now 8 isolates are flagged. In total, 79.3% of all isolates are marked ‘first weighted’ - 50.8% more than when using the CLSI guideline. In real life, this novel algorithm will yield 5-10% more isolates than the classic CLSI guideline.
+Instead of 2, now 9 isolates are flagged. In total, 79.5% of all isolates are marked ‘first weighted’ - 51% more than when using the CLSI guideline. In real life, this novel algorithm will yield 5-10% more isolates than the classic CLSI guideline.
As with filter_first_isolate()
, there’s a shortcut for this new algorithm too:
So we end up with 15,861 isolates for analysis.
+So we end up with 15,909 isolates for analysis.
We can remove unneeded columns:
@@ -786,7 +786,6 @@date | patient_id | hospital | @@ -803,44 +802,11 @@|||||||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1 | -2015-01-18 | -F9 | -Hospital B | -B_ESCHR_COL | -R | -S | -R | -S | -M | -Gram negative | -Escherichia | -coli | -TRUE | -||||||||||
3 | -2016-02-14 | -J4 | -Hospital A | -B_ESCHR_COL | -R | -I | -S | -S | -M | -Gram negative | -Escherichia | -coli | -TRUE | -||||||||||
4 | -2010-12-25 | -P2 | -Hospital B | +2011-10-07 | +O2 | +Hospital C | B_STRPT_PNE | -S | +R | S | S | R | @@ -851,51 +817,78 @@TRUE | ||||||||||
5 | -2016-12-26 | -S8 | +2013-10-20 | +C10 | Hospital A | -B_STRPT_PNE | -S | -I | +B_ESCHR_COL | S | R | +S | +S | +M | +Gram negative | +Escherichia | +coli | +TRUE | +|||||
2014-08-25 | +O4 | +Hospital A | +B_ESCHR_COL | +S | +S | +S | +S | F | +Gram negative | +Escherichia | +coli | +TRUE | +|||||||||||
2011-10-28 | +E3 | +Hospital B | +B_STRPT_PNE | +S | +S | +S | +R | +M | Gram positive | Streptococcus | pneumoniae | TRUE | |||||||||||
6 | -2010-03-27 | -R7 | -Hospital D | +2010-08-31 | +H1 | +Hospital A | B_KLBSL_PNE | R | S | +R | S | -S | -F | +M | Gram negative | Klebsiella | pneumoniae | TRUE | |||||
8 | -2016-08-08 | -K8 | -Hospital B | -B_KLBSL_PNE | +2014-11-03 | +U7 | +Hospital A | +B_STPHY_AUR | +S | +S | R | -I | S | -S | -M | -Gram negative | -Klebsiella | -pneumoniae | +F | +Gram positive | +Staphylococcus | +aureus | TRUE |
1 | Escherichia coli | -7,879 | -49.7% | -7,879 | -49.7% | +7,837 | +49.3% | +7,837 | +49.3% | ||||||||||||||
2 | Staphylococcus aureus | -3,915 | -24.7% | -11,794 | -74.4% | +3,940 | +24.8% | +11,777 | +74.0% | ||||||||||||||
3 | Streptococcus pneumoniae | -2,482 | -15.6% | -14,276 | -90.0% | +2,554 | +16.1% | +14,331 | +90.1% | ||||||||||||||
4 | Klebsiella pneumoniae | -1,585 | -10.0% | -15,861 | +1,578 | +9.9% | +15,909 | 100.0% | |||||||||||||||
Hospital A | -0.4759916 | +0.4717819 | |||||||||||||||||||||
Hospital B | -0.4808997 | +0.4802632 | |||||||||||||||||||||
Hospital C | -0.4682779 | +0.4870184 | |||||||||||||||||||||
Hospital D | -0.4651015 | +0.4804169 |
In the table above, all measurements are shown in milliseconds (thousands of seconds). A value of 5 milliseconds means it can determine 200 input values per second. It case of 100 milliseconds, this is only 10 input values per second. The second input is the only one that has to be looked up thoroughly. All the others are known codes (the first one is a WHONET code) or common laboratory codes, or common full organism names like the last one. Full organism names are always preferred.
To achieve this speed, the as.mo
function also takes into account the prevalence of human pathogenic microorganisms. The downside is of course that less prevalent microorganisms will be determined less fast. See this example for the ID of Thermus islandicus (B_THERMS_ISL
), a bug probably never found before in humans:
T.islandicus <- microbenchmark(as.mo("theisl"),
@@ -235,13 +235,13 @@
times = 10)
print(T.islandicus, unit = "ms", signif = 3)
#> Unit: milliseconds
-#> expr min lq mean median uq max neval
-#> as.mo("theisl") 444.0 449.0 479.0 488.0 493.0 506.0 10
-#> as.mo("THEISL") 444.0 484.0 488.0 491.0 507.0 514.0 10
-#> as.mo("T. islandicus") 80.5 80.8 87.8 81.3 89.9 118.0 10
-#> as.mo("T. islandicus") 79.8 80.4 82.0 80.7 81.2 93.5 10
-#> as.mo("Thermus islandicus") 63.4 63.5 72.3 64.0 64.5 107.0 10
That takes 7.7 times as much time on average. A value of 100 milliseconds means it can only determine ~10 different input values per second. We can conclude that looking up arbitrary codes of less prevalent microorganisms is the worst way to go, in terms of calculation performance. Full names (like Thermus islandicus) are almost fast - these are the most probable input from most data sets.
+#> expr min lq mean median uq max neval +#> as.mo("theisl") 287.0 296.0 329.0 330.0 334.0 432 10 +#> as.mo("THEISL") 286.0 292.0 333.0 329.0 366.0 433 10 +#> as.mo("T. islandicus") 72.9 73.1 90.1 75.7 94.1 161 10 +#> as.mo("T. islandicus") 72.8 73.5 89.4 79.7 115.0 125 10 +#> as.mo("Thermus islandicus") 65.8 66.0 76.8 67.7 85.2 107 10 +That takes 7.2 times as much time on average. A value of 100 milliseconds means it can only determine ~10 different input values per second. We can conclude that looking up arbitrary codes of less prevalent microorganisms is the worst way to go, in terms of calculation performance. Full names (like Thermus islandicus) are almost fast - these are the most probable input from most data sets.
In the figure below, we compare Escherichia coli (which is very common) with Prevotella brevis (which is moderately common) and with Thermus islandicus (which is very uncommon):
par(mar = c(5, 16, 4, 2)) # set more space for left margin text (16)
@@ -287,8 +287,8 @@
print(run_it, unit = "ms", signif = 3)
#> Unit: milliseconds
#> expr min lq mean median uq max neval
-#> mo_fullname(x) 743 771 805 798 844 886 10
So transforming 500,000 values (!!) of 50 unique values only takes 0.8 seconds (798 ms). You only lose time on your unique input values.
+#> mo_fullname(x) 716 738 778 763 778 899 10 +So transforming 500,000 values (!!) of 50 unique values only takes 0.76 seconds (762 ms). You only lose time on your unique input values.
So going from mo_fullname("Staphylococcus aureus")
to "Staphylococcus aureus"
takes 0.0005 seconds - it doesn’t even start calculating if the result would be the same as the expected resulting value. That goes for all helper functions:
So going from mo_fullname("Staphylococcus aureus")
to "Staphylococcus aureus"
takes 0.0006 seconds - it doesn’t even start calculating if the result would be the same as the expected resulting value. That goes for all helper functions:
run_it <- microbenchmark(A = mo_species("aureus"),
B = mo_genus("Staphylococcus"),
C = mo_fullname("Staphylococcus aureus"),
@@ -317,14 +317,14 @@
print(run_it, unit = "ms", signif = 3)
#> Unit: milliseconds
#> expr min lq mean median uq max neval
-#> A 0.330 0.399 0.444 0.425 0.480 0.599 10
-#> B 0.343 0.362 0.386 0.376 0.425 0.439 10
-#> C 0.327 0.454 0.550 0.571 0.640 0.816 10
-#> D 0.273 0.306 0.329 0.319 0.366 0.392 10
-#> E 0.246 0.266 0.295 0.286 0.323 0.364 10
-#> F 0.260 0.265 0.320 0.312 0.364 0.407 10
-#> G 0.238 0.252 0.281 0.270 0.319 0.339 10
-#> H 0.251 0.278 0.316 0.320 0.358 0.381 10
Of course, when running mo_phylum("Firmicutes")
the function has zero knowledge about the actual microorganism, namely S. aureus. But since the result would be "Firmicutes"
too, there is no point in calculating the result. And because this package ‘knows’ all phyla of all known bacteria (according to the Catalogue of Life), it can just return the initial value immediately.
Currently supported are German, Dutch, Spanish, Italian, French and Portuguese.
diff --git a/docs/articles/benchmarks_files/figure-html/unnamed-chunk-5-1.png b/docs/articles/benchmarks_files/figure-html/unnamed-chunk-5-1.png index e4fd0ba2..9e93bdfe 100644 Binary files a/docs/articles/benchmarks_files/figure-html/unnamed-chunk-5-1.png and b/docs/articles/benchmarks_files/figure-html/unnamed-chunk-5-1.png differ