In this validation we wanted to test if insertion was correctly implemented.
version: invadego 0.1.3
| Bias | SampleID | Seed |
|---|---|---|
| -100 | mb100 | 1687986419473565499 |
| -90 | mb90 | 1687986439519204740 |
| -80 | mb80 | 1687986459385010207 |
| -70 | mb70 | 1687986479004319218 |
| -60 | mb60 | 1687986498415164945 |
| -50 | mb50 | 1687986518191731481 |
| -40 | mb40 | 1687986537947039053 |
| -30 | mb30 | 1687986557979563694 |
| -20 | mb20 | 1687986577799895192 |
| -10 | mb10 | 1687986597553426300 |
| 0 | b0 | 1687986617358973799 |
| 10 | b10 | 1687986636894417837 |
| 20 | b20 | 1687986656418196570 |
| 30 | b30 | 1687986677115413538 |
| 40 | b40 | 1687986698520036068 |
| 50 | b50 | 1687986720075130951 |
| 60 | b60 | 1687986741333869430 |
| 70 | b70 | 1687986762991078587 |
| 80 | b80 | 1687986784961325051 |
| 90 | b90 | 1687986806594066852 |
| 100 | b100 | 1687986827704107589 |
tool="./main"
genome="mb:1,1,1,1,1"
cluster="kb:30,30,30,30,30"
rep=100
gen=1
steps=1
folder="Simulation-Results/Insertion-Bias/validation_7"
rr="0,0,0,0,0"
mkdir -p $folder
# Loop over values from -100 to 100 in steps of 10
for j in $(seq -100 10 100)
do
# Set basepop directly to "1000;j"
basepop="1000($j)"
# Assign current counter value to sampleid with descriptive prefix
if [ $j -ge 0 ]
then
sampleid="b${j}"
else
sampleid="mb${j#-}" # Use parameter expansion to remove the negative sign
fi
# Run the command and write the output to a file named after the sampleid
$tool --N 100000 --gen $gen --genome $genome --cluster $cluster --rr $rr --rep $rep --basepop "$basepop" --steps $steps --sampleid $sampleid > "$folder/result_${sampleid}.out"
done
cat result_*.out | grep -v "^Invade" | grep -v "^#" > combined_results.out# Load necessary libraries
library(readr)
library(dplyr)
library(ggplot2)
# Define column names
column_names <- c("rep", "gen", "popstat", "spacer_1", "fwte", "avw", "min_w", "avtes", "avpopfreq",
"fixed", "spacer_2", "phase", "fwcli", "avcli", "fixcli", "spacer_3",
"avbias", "3tot", "3cluster", "spacer_4", "sampleid")
# Load DataFrame with column names
df <- read_delim('Simulation-Results_Files/validation_7/combined_results.out', delim='\t', col_names = column_names)
# Define replacement dictionary
replace_dict <- c("mb100" = "-100","mb90" = "-90", "mb80" = "-80", "mb70" = "-70", "mb60" = "-60",
"mb50" = "-50", "mb40" = "-40", "mb30" = "-30", "mb20" = "-20",
"mb10" = "-10", "b100" = "100","b90" = "90", "b80" = "80", "b70" = "70",
"b60" = "60", "b50" = "50", "b40" = "40", "b30" = "30",
"b20" = "20", "b10" = "10", "b0" = "0")
# Apply replacements to 'sampleid' column
df$sampleid <- as.character(df$sampleid) %>% str_replace_all(replace_dict)
# Convert the columns to numeric
numeric_columns <- c("rep", "gen", "fwte", "avw", "min_w", "avtes", "avpopfreq",
"fixed", "fwcli", "avcli", "fixcli",
"avbias", "sampleid")
df[numeric_columns] <- lapply(df[numeric_columns], function(x) as.numeric(as.character(x)))
# Define your function
pc <- function(bias, clufrac) {
genfrac = 1.0 - clufrac
bias = bias / 100
clufit = (bias + 1.0) / 2.0
genfit = 1.0 - clufit
totfit = clufrac * clufit + genfrac * genfit
p = (clufrac * clufit) / totfit
return(p * 100)
}
# Filter dataframe where 'gen' == 0 and sort it by 'sampleid'
df2 <- df[df$gen == 0, numeric_columns] %>% arrange(sampleid)
# Calculate the expected values (pc) for each 'sampleid'
df2$pc <- sapply(df2$sampleid, function(x) pc(x, 0.03))# Define the common theme for consistency
common_theme <- function() {
theme_minimal() +
theme(
plot.title = element_text(hjust = 0.5, size = 14),
axis.title = element_text(size = 12),
axis.text = element_text(size = 10),
legend.position = "bottom",
panel.background = element_rect(fill = "white"), # Ensure white background
plot.background = element_rect(fill = "white", color = NA) # Remove panel border
)
}
# Create the plot with annotations for line and dot information
a <- ggplot(df2, aes(x = sampleid)) +
geom_point(aes(y = avcli), color = "#0072B2") + # Blue color for avcli points
geom_line(aes(y = pc), color = "black") + # Orange color for pc line
labs(
title = "Average Cluster Insertion Expected vs Observed Value Across Insertion Bias",
x = "Insertion Bias",
y = "Average Cluster Insertion"
) +
annotate("text", x = Inf, y = Inf, label = "Black Line: Expected\nBlue Points: Observed",
hjust = 1.5, vjust = 2, fontface = "italic", color = "black", size = 4,
position = position_nudge(y = -10)) + # Annotation positioned at the top right
common_theme()
# Save the plot with high resolution
ggsave(
filename = "images/Validation_7_1A.png",
plot = a,
width = 8,
height = 6,
units = "in",
dpi = 300 # Sufficient for high-quality prints
)
ggsave(
filename = "images/Validation_7_1A.pdf",
plot = a,
width = 8,
height = 6,
units = "in",
dpi = 300 # Sufficient for high-quality prints
)
The distribution of average TE (Transposable Elements) insertions across different insertion bias levels for all replicates. The x-axis shows the different Insertion Bias levels ranging from -100 to 100. The y-axis represents the average TE insertions in the piRNA Cluster (Blue Dots for Observed Value and Orange line for Expected value) for each bias level.
# Filter out the group with sampleid = 100 and -100
df_filtered <- df2 %>%
filter(!(sampleid %in% c(-100, 100)))
# Subtract 'pc' from 'avcli' and force as numeric
df_filtered$TE_insertions <- as.numeric(as.character(df_filtered$avcli)) - as.numeric(as.character(df_filtered$pc))
df_filtered$TE_insertions <- 100 * df_filtered$TE_insertions
# Convert 'sampleid' to factor
df_filtered$insertion_bias <- as.factor(df_filtered$sampleid)
# Create a new plot file with larger dimensions
png(filename = "images/Validation_7_1B.jpg", width = 800, height = 600)
# Boxplot
boxplot(df_filtered$TE_insertions ~ df_filtered$insertion_bias,
border = rgb(0.1, 0.1, 0.7, 0.5),
main = "Variability in Cluster Insertion Across 100 Replications",
xlab = "Insertion Bias",
ylab = "TE Cluster Insertion Variability Across 100 Replications")
# Add data points
mylevels <- levels(df_filtered$insertion_bias)
levelProportions <- summary(df_filtered$insertion_bias)/nrow(df_filtered)
for(i in 1:length(mylevels)){
thislevel <- mylevels[i]
thisvalues <- df_filtered[df_filtered$insertion_bias==thislevel, "TE_insertions"]
# Ensure thisvalues is a numeric vector
thisvalues <- unlist(thisvalues)
# Take the x-axis indices and add a jitter, proportional to the N in each level
myjitter <- jitter(rep(i, length(thisvalues)), amount=levelProportions[i]/2)
# Use smaller points
points(myjitter, thisvalues, pch=20, cex=0.9, col=rgb(0,0.2,0.25,0.6))
}
# Close the plot file
invisible(dev.off())
The plot displays the variability in Transposable Elements (TE) insertions across different levels of Insertion Bias, using data from 100 replications. To calculate the ‘TE_insertions’ values plotted on the y-axis, the ‘avcli’ column was subtracted from the ‘pc’ column in the original dataset. This difference was then multiplied by 100 to express the relative difference as a percentage:
The ‘Insertion Bias’ variable, represented on the x-axis, corresponds to the ‘sampleid’ column. It was converted to a factor variable to accommodate the categorical nature of this variable. The boxplot summarizes the distribution of ‘TE_insertions’ for each level of ‘Insertion Bias’. Each boxplot indicates the median, interquartile range, and potential outliers. The individual data points overlaid on the boxplots represent the variability in TE insertions.
# Calculate mean_cli, sd_meancli, pc, and deviation_pc
df_summary <- df2 %>%
group_by(sampleid) %>%
summarise(mean_cli = mean(avcli, na.rm = TRUE),
sd_meancli = sd(avcli, na.rm = TRUE)) %>%
mutate(pc = sapply(sampleid, function(x) pc(x, 0.03)),
deviation_pc = mean_cli - pc)
# Create scatter plot with error bars
c <- ggplot(df_summary, aes(x = sampleid)) +
geom_point(aes(y = mean_cli), color = "#003f5c") +
geom_errorbar(aes(ymin = mean_cli - sd_meancli, ymax = mean_cli + sd_meancli), width = .2, color = "003f5c") +
geom_line(aes(y = pc), color = "#ffa600") +
labs(title = "Observed and Expected Average Cluster Insertion across Insertion Bias",
x = "Insertion Bias",
y = "Mean of Average Cluster Insertion Observed (Dots)") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5))
ggsave(filename = "images/Validation_7_1C.jpg", plot = c, width = 10, height = 10, units = "in")
ggsave(filename = "images/Validation_7_1C.pdf", plot = c, width = 10, height = 10, units = "in", device = "pdf")
This plot visualizes the mean of Average
Cluster Insertion across various Insertion Bias levels, with error bars
indicating standard deviation. Unlike previous plot’s individual data
points, this representation provides a concise summary of the data
(mean), making it easier to discern overall trends and reducing the
impact of outliers. The expected values are depicted by a dashed red
line.
# Transform the 'mean_cli' and 'sd_meancli' by taking log10 to visualize better.
# We add a small constant to avoid log(0) and handle negative values.
# Apply a logarithmic transformation to 'mean_cli', 'sd_meancli', and 'pc'
df_summary$log_mean_cli <- log10(df_summary$mean_cli + 0.0001)
df_summary$log_sd_meancli <- log10(df_summary$sd_meancli + 0.0001)
df_summary$log_pc <- log10(df_summary$pc + 0.0001)
# Create scatter plot with error bars
d <- ggplot(df_summary, aes(x = sampleid)) +
geom_point(aes(y = log_mean_cli), color = "blue") +
geom_errorbar(aes(ymin = log_mean_cli - log_sd_meancli, ymax = log_mean_cli + log_sd_meancli), width = .2, color = "#008080") +
geom_line(aes(y = log_pc), color = "orange") +
labs(title = "Observed and Expected Average Cluster Insertion across Insertion Bias",
x = "Insertion Bias",
y = "Log-transformed Mean of Average Cluster Insertion") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5))
# Print the plot
ggsave(filename = "images/Validation_7_1D.jpg", plot = d, width = 10, height = 10, units = "in")
ggsave(filename = "images/Validation_7_1D.pdf", plot = d, width = 10, height = 10, units = "in")
Figure 3 presents the same data as Figure 2, but with a logarithmic transformation applied to the y-values. The transformation is performed to better visualize the standard deviation, which is relatively small compared to the mean values. By using the log scale, we can more easily see the variability in the data (shown by the error bars) and the deviation of the observed values (blue points) from the expected values (red line). This plot provides a more detailed understanding of the distribution of the data and the nature of the discrepancies between the observed and expected values.
# Filter for rep 1
df_rep1 <- df2[df2$rep == 1, ]
# Create scatter plot
e <- ggplot(df_rep1, aes(x = sampleid, y = avtes)) +
geom_point(color = "#003f5c") +
labs(title = "Average Transposable Element Insertions",
x = "Insertion Bias",
y = "Average TE Insertions") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5))
ggsave(filename = "images/Validation_7_1E.jpg", plot = e, width = 10, height = 10, units = "in")
ggsave(filename = "images/Validation_7_1E.pdf", plot = e, width = 10, height = 10, units = "in")
Average TE insertions (y-axis) and insertion bias (x-axis) for a single replicate . The x-axis represents the Insertion Bias varying from -100 to 100. The y-axis depicts the average TE insertions for each bias level. Each data point on the graph represents the average TE insertions for a specific bias level for replicate 1.
# Calculate mean and standard deviation of avtes
df_summary_2 <- df2 %>%
group_by(sampleid) %>%
summarise(mean_avtes = mean(avtes, na.rm = TRUE),
sd_avtes = sd(avtes, na.rm = TRUE))
# Create scatter plot with error bars
f <- ggplot(df_summary_2, aes(x = sampleid)) +
geom_point(aes(y = mean_avtes), color = "blue") +
geom_errorbar(aes(ymin = mean_avtes - sd_avtes, ymax = mean_avtes + sd_avtes), width = .2, color = "#008080") +
labs(title = "Average TE Insertions across Insertion Bias",
x = "Insertion Bias",
y = "Average TE Insertions") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5))
ggsave(filename = "images/Validation_7_1F.jpg", plot = f, width = 10, height = 10, units = "in")
ggsave(filename = "images/Validation_7_1F.pdf", plot = f, width = 10, height = 10, units = "in")
All replicates’ mean average TE insertions (y-axis) against the insertion bias (x-axis). The plot includes error bars, which visually represent the variability in the data by displaying the standard deviation for the average TE insertions at each bias level. The x-axis represents the Insertion Bias, which ranges from -100 to 100. The y-axis depicts the mean of the average TE insertions for each bias level. Each data point on the graph represents the mean average TE insertions for a specific bias level across all replicates, with the error bars representing the standard deviation.
Across the 21 insertion bias values (sampleid 0 to 20), the mean of cli is calculated for all 100 replications. The standard deviation is also calculated. We have another column named pc which is the theoretical value of average cluster insertion for a given insertion bias with a 3% cluster size. It was calculated as follows:
# Define your function
pc <- function(bias, clufrac) {
genfrac <- 1.0 - clufrac
bias <- bias / 100
clufit <- (bias + 1.0) / 2.0
genfit <- 1.0 - clufit
totfit <- clufrac * clufit + genfrac * genfit
p <- (clufrac * clufit) / totfit
return(p * 100)
}We create a new dataframe where ‘gen’ equals 0 and sort it by ‘sampleid’. Then we calculate the expected values (pc) for each ‘sampleid’.
# Filter dataframe where 'gen' equals 0 and sort it by 'sampleid'
df2 <- df[df$gen == 0, numeric_columns] %>% arrange(sampleid)
# Calculate the expected values (pc) for each 'sampleid'
df2$pc <- sapply(df2$sampleid, function(x) pc(x, 0.03))The last column, deviation_pc, is calculated as the deviation of mean_cli from pc as follows:
# Calculate mean_cli, sd_meancli, pc, and deviation_pc
df_summary <- df2 %>%
group_by(sampleid) %>%
summarise(mean_cli = mean(avcli, na.rm = TRUE),
sd_meancli = sd(avcli, na.rm = TRUE)) %>%
mutate(pc = sapply(sampleid, function(x) pc(x, 0.03)),
deviation_pc = mean_cli - pc)
print(tibble(df_summary), n=21)## # A tibble: 21 × 5
## sampleid mean_cli sd_meancli pc deviation_pc
## <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 -100 0 0 0 0
## 2 -90 0.162 0.0130 0.163 -0.000814
## 3 -80 0.341 0.0204 0.342 -0.00157
## 4 -70 0.542 0.0248 0.543 -0.000423
## 5 -60 0.769 0.0297 0.767 0.00154
## 6 -50 1.02 0.0307 1.02 -0.00111
## 7 -40 1.30 0.0360 1.31 -0.00454
## 8 -30 1.63 0.0417 1.64 -0.00397
## 9 -20 2.02 0.0448 2.02 -0.00160
## 10 -10 2.48 0.0459 2.47 0.00719
## 11 0 2.99 0.0514 3 -0.00930
## 12 10 3.64 0.0571 3.64 0.000316
## 13 20 4.43 0.0622 4.43 -0.00360
## 14 30 5.44 0.0707 5.43 0.00955
## 15 40 6.73 0.0861 6.73 0.00333
## 16 50 8.50 0.0932 8.49 0.00513
## 17 60 11.0 0.0994 11.0 -0.00867
## 18 70 14.9 0.112 14.9 -0.0104
## 19 80 21.8 0.144 21.8 0.00271
## 20 90 37.0 0.164 37.0 0.0114
## 21 100 100. 0.00485 100 -0.0174
The validation matches our expectations and the insertion is working as expected and the the simulation successfully incorporates the user-defined TE insertions as specified.