No negative values in my data but plot shows negative

ghz 8months ago ⋅ 95 views

There are no negative values in my data but the violin plot shows negative values. I have over 60,000 rows of data (2 columns, col 1 is subgenome and col 2 is value) to plot. Please suggest how can I share such massive data? Below is my code. Please suggest.

library(ggplot2)
library(Hmisc)

# Read the data
df <- read.table('ABCD-meth-r1.tsv', header = TRUE, sep = "\t")

# Plotting code
p <- ggplot(df, aes(x = subgenome, y = value, fill = subgenome)) +
geom_violin(trim = FALSE) + 
stat_summary(fun.data="mean_sdl", fun.args = list(mult = 2), geom="crossbar", 
width=0.2) +
stat_summary(fun.data="mean_sdl", fun.args = list(mult = 2), geom="pointrange", 
color="black") +
xlab("") +
ylab("CDS methylation") +
theme_bw()

# Print the plot
print(p)

output of my code: enter image description here

with trim = TRUE enter image description here

generated withOUT using stat_summary: but still see plot going below 0. Please help me understand this? enter image description here

Answers

The appearance of negative values in your violin plot may be due to outliers or extreme values in your data. Violin plots display the distribution of the data, and if there are outliers, the tails of the violin plot can extend beyond the range of the actual data.

Here are a few suggestions to address this issue and effectively visualize your data:

  1. Inspect your data: Check your dataset for any unexpected negative values. It's possible that there are errors or anomalies in your data that need to be addressed.

  2. Use log scale: If your data contains highly skewed distributions or large ranges, you may consider using a log scale for the y-axis. This can help in visualizing the data more effectively, especially when there are extreme values.

    p <- ggplot(df, aes(x = subgenome, y = value, fill = subgenome)) +
     geom_violin(trim = FALSE) + 
     scale_y_log10() +  # Use log scale for y-axis
     xlab("") +
     ylab("CDS methylation") +
     theme_bw()

  3. Trim the outliers: You can trim the outliers from your data before plotting to prevent them from affecting the appearance of the violin plot. This can be done using statistical methods or by specifying a specific range of values to include.

  4. Adjust the bandwidth: The bandwidth parameter in geom_violin() controls the smoothness of the violin plot. You can experiment with different bandwidth values to see if it affects the appearance of the plot.

  5. Consider other visualization techniques: If violin plots are not suitable for your data due to outliers or other issues, consider using other visualization techniques such as boxplots, histograms, or density plots.

  6. Subset your data: Since you have a large dataset, you may consider subsetting it to a smaller sample size for visualization purposes. This can help in improving the plot's readability and performance.

By applying these techniques and carefully examining your data, you should be able to create a violin plot that effectively visualizes the distribution of your data without negative values or misleading artifacts.

#r