From 25d6b9be33a2bb450aec37fdf74242d08f0cecfd Mon Sep 17 00:00:00 2001
From: james94 <james94@mi.fu-berlin.de>
Date: Fri, 5 Aug 2022 14:45:16 +0200
Subject: [PATCH] rules/build_hifi.smk
update hifi build
---
scripts/addFullTableForReport.py | 23 +--
scripts/colouredHeatmap_legend.tsv | 5 +
scripts/fullTable_heatmap_external.R | 132 +++++++++++++++++
.../fullTable_heatmap_internalComparison.R | 140 ++++++++++++++++++
scripts/internalComparison_legend.tsv | 3 +
scripts/tableOnSamePage.css | 3 +
6 files changed, 296 insertions(+), 10 deletions(-)
create mode 100644 scripts/colouredHeatmap_legend.tsv
create mode 100644 scripts/fullTable_heatmap_external.R
create mode 100644 scripts/fullTable_heatmap_internalComparison.R
create mode 100644 scripts/internalComparison_legend.tsv
create mode 100644 scripts/tableOnSamePage.css
diff --git a/scripts/addFullTableForReport.py b/scripts/addFullTableForReport.py
index bb37771..6ac763b 100644
--- a/scripts/addFullTableForReport.py
+++ b/scripts/addFullTableForReport.py
@@ -2,8 +2,8 @@ import pandas as pd
from tabulate import tabulate
samples=pd.read_csv(snakemake.input.results, dtype=str, index_col=0, delim_whitespace=True, skip_blank_lines=True)
# samples=samples.reset_index
-turn2FloatAndMb=['Bases_Mb','Est_Size_Mb','Longest_Scaff_Mb','Scaff_N50_Mb','Scaff_NG50_Mb','Scaff_N95_Mb','Scaff_NG95_Mb','Longest_Cont','Cont_N50_Mb','Cont_NG50_Mb','Cont_N95_Mb','Cont_NG95_Mb']
-roundDecimals=['Comp_BUSCOs_%','Comp_Single_BUSCOs_%','Het_%','GC_%','QV','Completeness','Bases_Mb','Est_Size_Mb','Longest_Scaff_Mb','Scaff_N50_Mb','Scaff_NG50_Mb','Scaff_N95_Mb','Scaff_NG95_Mb','Longest_Cont','Cont_N50_Mb','Cont_NG50_Mb','Cont_N95_Mb','Cont_NG95_Mb']
+turn2FloatAndMb=['Bases_Mb','Est_Size_Mb','Longest_Scaff_Mb','Scaff_N50_Mb','Scaff_NG50_Mb','Scaff_N95_Mb','Scaff_NG95_Mb','Longest_Cont_Mb','Cont_N50_Mb','Cont_NG50_Mb','Cont_N95_Mb','Cont_NG95_Mb']
+roundDecimals=['Comp_BUSCOs_%','Comp_Single_BUSCOs_%','Het_%','GC_%','QV','Completeness','Bases_Mb','Est_Size_Mb','Longest_Scaff_Mb','Scaff_N50_Mb','Scaff_NG50_Mb','Scaff_N95_Mb','Scaff_NG95_Mb','Longest_Cont_Mb','Cont_N50_Mb','Cont_NG50_Mb','Cont_N95_Mb','Cont_NG95_Mb']
print('this is samples',samples)
sampleTransposed=samples.T
print('this is sampleTransposed',sampleTransposed)
@@ -13,11 +13,14 @@ for i in range(0,4):
sampleTransposed[roundDecimals[i]]=sampleTransposed[roundDecimals[i]].str.replace('%','')
for roundHeader in roundDecimals:
sampleTransposed[roundHeader]=sampleTransposed[roundHeader].astype(float).round(2)
-with open(snakemake.output[0], "w") as out:
- print("", file=out)
- print("\\blandscape", file=out)
- print("", file=out)
- print("\\tiny", file=out)
- print(tabulate(sampleTransposed.rename_axis('ASM_ID'), headers='keys',tablefmt="pipe", showindex=True), file=out)
- print("\\elandscape", file=out)
- print("", file=out)
+with open(snakemake.output[0], "w") as out_markdown:
+ print("", file=out_markdown)
+ print("\\blandscape", file=out_markdown)
+ print("", file=out_markdown)
+ print("\\tiny", file=out_markdown)
+ print(tabulate(sampleTransposed.rename_axis('ASM_ID'), headers='keys',tablefmt="pipe", showindex=True), file=out_markdown)
+ print("\\elandscape", file=out_markdown)
+ print("", file=out_markdown)
+
+with open(snakemake.output[1], "w") as out_plain:
+ print(tabulate(sampleTransposed.rename_axis('ASM_ID'), headers='keys',tablefmt="plain", showindex=True), file=out_plain)
diff --git a/scripts/colouredHeatmap_legend.tsv b/scripts/colouredHeatmap_legend.tsv
new file mode 100644
index 0000000..526c788
--- /dev/null
+++ b/scripts/colouredHeatmap_legend.tsv
@@ -0,0 +1,5 @@
+Gaps_per_Gb Scaff_NG50_Mb Cont_NG50_Mb QV Completeness Comp_Single_BUSCOs_%
+'< 200' '> 100Mbp' '> 10Mbp' '> 50' '> 95%' '> 95%'
+'200 - 1000' '10Mbp - 100Mbp' '1Mbp - 10Mbp' '40 - 50' '90% - 95%' '90% - 95%'
+'1000 - 10000' '0.1Mbp - 10Mbp' '0.01Mbp - 1Mbp' '35 - 40' '80% - 90%' '80% - 90%'
+'> 10000' '< 0.1Mbp' '< 0.01Mbp' '< 35' '< 80%' '< 80%'
diff --git a/scripts/fullTable_heatmap_external.R b/scripts/fullTable_heatmap_external.R
new file mode 100644
index 0000000..3d251c8
--- /dev/null
+++ b/scripts/fullTable_heatmap_external.R
@@ -0,0 +1,132 @@
+suppressMessages(library(dplyr))
+suppressMessages(library(formattable))
+#suppressMessages(library(data.table))
+
+
+fullTableOfStats<-read.table(file = snakemake@input[[1]], sep = "", header = TRUE, row.names=NULL, check.names = FALSE)
+
+
+
+fullTableOfStats$Gaps_per_Gb <- ((fullTableOfStats$Gaps * fullTableOfStats$Est_Size_Mb)/1000)
+fullTableOfStats$Gaps_per_Gb <- as.integer(fullTableOfStats$Gaps_per_Gb)
+
+
+customPurple='#c699e8'
+customGreen='#8fc773'
+
+selectionOfStats_colouredHeatmap <- fullTableOfStats %>%
+ select(c('ASM_ID','ASM_LEVEL',
+ 'Gaps_per_Gb', 'Scaff_NG50_Mb',
+ 'Cont_NG50_Mb','QV',
+ 'Completeness', 'Comp_Single_BUSCOs_%'))
+
+sink(file = snakemake@output[[1]])
+format_table(selectionOfStats_colouredHeatmap,
+ align =c("l","c","c","c","c", "c", "c", "c", "c"),
+ list(Gaps_per_Gb = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ `background-color` = ifelse(ASM_LEVEL == 'cont', "#666666",
+ ifelse(between(Gaps_per_Gb,1001, 10000), "#FFCC99",
+ ifelse(between(Gaps_per_Gb,201 , 1000), customGreen,
+ ifelse(Gaps_per_Gb <= 200, customPurple, "#FF9999")))))),
+ Scaff_NG50_Mb = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ `background-color` = ifelse(between(Scaff_NG50_Mb,0.1, 9.99999), "#FFCC99",
+ ifelse(between(Scaff_NG50_Mb,10, 99.99999), customGreen,
+ ifelse(Scaff_NG50_Mb >= 100, customPurple, "#FF9999"))))),
+ Cont_NG50_Mb = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ `background-color` = ifelse(between(Cont_NG50_Mb,0.01, 0.99999), "#FFCC99",
+ ifelse(between(Cont_NG50_Mb,1, 9.999999), customGreen,
+ ifelse(Cont_NG50_Mb >= 10, customPurple, "#FF9999"))))),
+ Completeness = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ `background-color` = ifelse(between(Completeness,80, 89.9999), "#FFCC99",
+ ifelse(between(Completeness,90, 94.99999), customGreen,
+ ifelse(Completeness >= 95, customPurple, "#FF9999"))))),
+ `Comp_Single_BUSCOs_%` = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ `background-color` = ifelse(between(`Comp_Single_BUSCOs_%`,80, 89.9999), "#FFCC99",
+ ifelse(between(`Comp_Single_BUSCOs_%`,90, 95), customGreen,
+ ifelse(`Comp_Single_BUSCOs_%` >= 95, customPurple, "#FF9999"))))),
+ QV = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ `background-color` = ifelse(between(QV,35, 39.9999999999), "#FFCC99",
+ ifelse(between(QV,40, 49.999999999), customGreen,
+ ifelse(QV >= 50, customPurple, "#FF9999")))))))
+
+cat('<br>')
+cat("\n")
+cat('<br>')
+
+legendTable<-read.table(file = snakemake@params[[1]], sep = "", header = TRUE, row.names=NULL, check.names = FALSE)
+
+format_table(legendTable,
+ align =c("c","c","c", "c", "c", "c", "c"),
+ list(Gaps_per_Gb = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ "font.size" = "12px",
+ "width" = '150px',
+ `background-color` = ifelse(Gaps_per_Gb == '> 10000',"#FF9999",
+ ifelse(Gaps_per_Gb == '1000 - 10000',"#FFCC99",
+ ifelse(Gaps_per_Gb == '200 - 1000',customGreen, customPurple))))),
+ Scaff_NG50_Mb = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ "font.size" = "12px",
+ "width" = '150px',
+ `background-color` = ifelse(Scaff_NG50_Mb == '< 0.1Mbp',"#FF9999",
+ ifelse(Scaff_NG50_Mb == '0.1Mbp - 10Mbp',"#FFCC99",
+ ifelse(Scaff_NG50_Mb == '10Mbp - 100Mbp',customGreen, customPurple))))),
+ Cont_NG50_Mb = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ "font.size" = "12px",
+ "width" = '150px',
+ `background-color` = ifelse(Cont_NG50_Mb == '< 0.01Mbp',"#FF9999",
+ ifelse(Cont_NG50_Mb == '0.01Mbp - 1Mbp',"#FFCC99",
+ ifelse(Cont_NG50_Mb == '1Mbp - 10Mbp',customGreen, customPurple))))),
+ QV = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ "font.size" = "12px",
+ "width" = '100px',
+ `background-color` = ifelse(QV == '< 35',"#FF9999",
+ ifelse(QV == '35 - 40',"#FFCC99",
+ ifelse(QV == '40 - 50',customGreen, customPurple))))),
+ Completeness = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ "font.size" = "12px",
+ "width" = '150px',
+ `background-color` = ifelse(Completeness == '< 80%',"#FF9999",
+ ifelse(Completeness == '80% - 90%',"#FFCC99",
+ ifelse(Completeness == '90% - 95%',customGreen, customPurple))))),
+ `Comp_Single_BUSCOs_%` = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ "font.size" = "12px",
+ "width" = '220px',
+ `background-color` = ifelse(`Comp_Single_BUSCOs_%` == '< 80%',"#FF9999",
+ ifelse(`Comp_Single_BUSCOs_%` == '80% - 90%',"#FFCC99",
+ ifelse(`Comp_Single_BUSCOs_%` == '90% - 95%',customGreen, customPurple)))))))
+sink(file = NULL)
diff --git a/scripts/fullTable_heatmap_internalComparison.R b/scripts/fullTable_heatmap_internalComparison.R
new file mode 100644
index 0000000..e1d5ae7
--- /dev/null
+++ b/scripts/fullTable_heatmap_internalComparison.R
@@ -0,0 +1,140 @@
+suppressMessages(library(dplyr))
+suppressMessages(library(formattable))
+
+fullTableOfStats<-read.table(file = snakemake@input[[1]], sep = "", header = TRUE, row.names=NULL, check.names = FALSE)
+
+fullTableOfStats$Gaps_per_Gb <- ((fullTableOfStats$Gaps * fullTableOfStats$Est_Size_Mb)/1000)
+fullTableOfStats$Gaps_per_Gb <- as.integer(fullTableOfStats$Gaps_per_Gb)
+
+
+
+selectionOfStats_internalComparisonHeatmap <- fullTableOfStats %>%
+ select(c('ASM_ID','ASM_LEVEL', 'Bases_Mb', 'Het_%','GC_%', 'Gaps_per_Gb', 'Scaff', 'Cont',
+ 'Longest_Scaff_Mb','Scaff_NG50_Mb', 'Scaff_NG95_Mb',
+ 'Longest_Cont_Mb', 'Cont_NG50_Mb','Cont_NG95_Mb',
+ 'QV', 'Completeness',
+ 'Comp_BUSCOs_%','Comp_Single_BUSCOs_%'))
+
+
+
+
+customBlue_max = "#2e96ff"
+
+customBlue_min = "#dcecfc"
+
+
+customGray_max = "#8c8c8c"
+
+customGray_min = "#e6e6e6"
+
+sink(file = snakemake@output[[1]])
+
+format_table(selectionOfStats_internalComparisonHeatmap,
+ align =c("l","c","c","c","c", "c", "c", "c", "c"),
+ list(Bases_Mb = color_tile(customGray_min, customGray_max),
+ `Het_%` = color_tile(customGray_min, customGray_max),
+ `GC_%` = color_tile(customGray_min, customGray_max),
+ Gaps_per_Gb = color_tile(customBlue_max, customBlue_min),
+ Scaff = color_tile(customBlue_max,customBlue_min),
+ Cont = color_tile(customBlue_max,customBlue_min),
+ Longest_Scaff_Mb = color_tile(customBlue_min, customBlue_max),
+ Scaff_NG50_Mb = color_tile(customBlue_min, customBlue_max),
+ Scaff_NG95_Mb = color_tile(customBlue_min, customBlue_max),
+ Longest_Cont_Mb = color_tile(customBlue_min, customBlue_max),
+ Cont_NG50_Mb = color_tile(customBlue_min, customBlue_max),
+ Cont_NG95_Mb = color_tile(customBlue_min, customBlue_max),
+ QV = color_tile(customBlue_min, customBlue_max),
+ Completeness = color_tile(customBlue_min, customBlue_max),
+ `Comp_BUSCOs_%` = color_tile(customBlue_min, customBlue_max),
+ `Comp_Single_BUSCOs_%` = color_tile(customBlue_min, customBlue_max)))
+
+cat('<br>')
+cat("\n")
+cat('<br>')
+
+legendTable<-read.table(file = snakemake@params[[1]], sep = "", header = TRUE, row.names=NULL, check.names = FALSE)
+
+format_table(legendTable,
+ align =c("c","c","c", "c", "c", "c", "c","c","c","c", "c", "c", "c", "c","c", "c"),
+ list(Bases_Mb = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ `background-color` = ifelse(Bases_Mb == 'Max',customGray_max, customGray_min))),
+ `Het_%` = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ `background-color` = ifelse(`Het_%` == 'Max',customGray_max, customGray_min))),
+ `GC_%` = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ `background-color` = ifelse(`GC_%` == 'Max',customGray_max, customGray_min))),
+ Gaps_per_Gb = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ `background-color` = ifelse(Gaps_per_Gb == 'Min',customBlue_max, customBlue_min))),
+ Scaff = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ `background-color` = ifelse(Scaff == 'Min',customBlue_max, customBlue_min))),
+ Cont = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ `background-color` = ifelse(Cont == 'Min',customBlue_max, customBlue_min))),
+ Longest_Scaff_Mb = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ `background-color` = ifelse(Longest_Scaff_Mb == 'Max',customBlue_max, customBlue_min))),
+ Scaff_NG50_Mb = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ `background-color` = ifelse(Scaff_NG50_Mb == 'Max',customBlue_max, customBlue_min))),
+ Scaff_NG95_Mb = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ `background-color` = ifelse(Scaff_NG95_Mb == 'Max',customBlue_max, customBlue_min))),
+ Longest_Cont_Mb = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ `background-color` = ifelse(Longest_Cont_Mb == 'Max',customBlue_max, customBlue_min))),
+ Cont_NG50_Mb = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ `background-color` = ifelse(Cont_NG50_Mb == 'Max',customBlue_max, customBlue_min))),
+ Cont_NG95_Mb = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ `background-color` = ifelse(Cont_NG95_Mb == 'Max',customBlue_max, customBlue_min))),
+ QV = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ `background-color` = ifelse(QV == 'Max',customBlue_max, customBlue_min))),
+ Completeness = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ `background-color` = ifelse(Completeness == 'Max',customBlue_max, customBlue_min))),
+ `Comp_BUSCOs_%` = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ `background-color` = ifelse(`Comp_BUSCOs_%` == 'Max',customBlue_max, customBlue_min))),
+ `Comp_Single_BUSCOs_%` = formatter("span",
+ style = ~style(display = "block",
+ padding = "0 4px",
+ `border-radius` = "3px",
+ `background-color` = ifelse(`Comp_Single_BUSCOs_%` == 'Max',customBlue_max, customBlue_min)))))
+
+sink(file = NULL)
diff --git a/scripts/internalComparison_legend.tsv b/scripts/internalComparison_legend.tsv
new file mode 100644
index 0000000..b8e16cc
--- /dev/null
+++ b/scripts/internalComparison_legend.tsv
@@ -0,0 +1,3 @@
+Bases_Mb Het_% GC_% Gaps_per_Gb Scaff Cont Longest_Scaff_Mb Scaff_NG50_Mb Scaff_NG95_Mb Longest_Cont_Mb Cont_NG50_Mb Cont_NG95_Mb QV Completeness Comp_BUSCOs_% Comp_Single_BUSCOs_%
+'Max' 'Max' 'Max' 'Max' 'Max' 'Max' 'Max' 'Max' 'Max' 'Max' 'Max' 'Max' 'Max' 'Max' 'Max' 'Max'
+'Min' 'Min' 'Min' 'Min' 'Min' 'Min' 'Min' 'Min' 'Min' 'Min' 'Min' 'Min' 'Min' 'Min' 'Min' 'Min'
diff --git a/scripts/tableOnSamePage.css b/scripts/tableOnSamePage.css
new file mode 100644
index 0000000..ad4e7ad
--- /dev/null
+++ b/scripts/tableOnSamePage.css
@@ -0,0 +1,3 @@
+<style>
+.main-container { width: 1200px; max-width:2800px;}
+</style>
--
GitLab