Sliding window plot using Python

Question

[A_Lh]

I want to plot the number of positions in a sliding window of 1000 and a step of 20 for each sample (A-D).

Interpretation:

• 1: position exists;
• NA: position does not exist.

I have tested a dozen tools in bash, R and other but I am looking for a Python solution.
Your advice please.

#This is an example of my data:
window = 1000
step = 20

# Example of dataframe
POSITION        A       B       C       D
1250            1       1       1       1 
1750            NA      1       NA      1
1786            1       NA      1       1
1812            1       1       1       1
1855            1       1       1       1
1896            1       NA      1       NA
2635            NA      1       1       1
1689            1       1       NA      NA
3250            1       1       1       1
3655            1       NA      1       1
3589            NA      1       1       1

I am looking for some thing like this:

Any help will be appreciated!

[Alex Reynolds]

Previous question: How to plot SNPs distribution on each chromosome?

Just need to adjust BEDOPS bedmap commands to count SNPs over sliding windows, and then feed that as input to the provided script.

Answer 1

[A_Lh]

This is a Python solution:

import pandas as pd
import numpy as np
import seaborn as sns
df = pd.DataFrame({'POSITION': [1250,
  1750,
  1786,
  1812,
  1855,
  1896,
  2635,
  1689,
  3250,
  3655,
  3589],
 'A': [1.0, np.nan, 1.0, 1.0, 1.0, 1.0, np.nan, 1.0, 1.0, 1.0, np.nan],
 'B': [1.0, 1.0, np.nan, 1.0, 1.0, np.nan, 1.0, 1.0, 1.0, np.nan, 1.0],
 'C': [1.0, np.nan, 1.0, 1.0, 1.0, 1.0, 1.0, np.nan, 1.0, 1.0, 1.0],
 'D': [1.0, 1.0, 1.0, 1.0, 1.0, np.nan, 1.0, np.nan, 1.0, 1.0, 1.0]})



window = 5
step = 2

df = df.set_index('POSITION').rolling(window).count().reset_index().iloc[::step, :]    
df = df.melt(id_vars='POSITION', value_vars=['A','B','C','D'], value_name='polym', var_name='chromop')    
sns.lineplot(data=df, x='POSITION',y='polym',hue='chromop')

Answer 2

[4galaxy77]

In R instead of python, but it will do the job:

library(zoo)
library(tidyverse)
count_positions = function(x) sum(x, na.rm=T)
as.data.frame(rollapply(dat[,-1], FUN=count_positions, width=2, by=2)) %>% 
    mutate(index = 1:n()) %>% 
    pivot_longer(-index) %>% 
    ggplot(aes(x=index, y=value, colour=name)) + 
    geom_line()