CODEWITHSUNDEEP
pythonbash
GenomeBio
GenomeBio

Reputation: 69

How to extract row counts based on multiple descriptors in a columns from csv and then export output a new csv using bash/python script?

I am working with a csv file (100s of rows) containing data as follows. I would like to get counts per each gene for each element in csv/tab format.

Input

    Gene     Element   
 ---------- ---------- 
  STBZIP1    G-box     
  STBZIP1    G-box     
  STBZIP1    MYC       
  STBZIP1    MYC       
  STBZIP1    MYC       
  STBZIP10   MYC       
  STBZIP10   MYC       
  STBZIP10   MYC       
  STBZIP10   G-box     
  STBZIP10   G-box     
  STBZIP10   G-box     
  STBZIP10   G-box

Expected output

    Gene     G-Box   MYC  
 ---------- ------- ----- 
  STBZIP1        2     3  
  STBZIP10       4     3

Can someone please help me to come up with a bash script (or python) in this regard?

Update

I am trying the following and stuck for the time being :| ;

import pandas as pd
df = pd.read_csv("Promoter_Element_Distribution.csv")
print (df)
df.groupby(['Gene', 'Element']).size().unstack(fill_value=0)

Upvotes: 1

Views: 90

Answers (2)

Amessihel
Amessihel

Reputation: 6414

Since you asked also for a bash version, here is a use of awk1. It's commented, and also the output is "well" formatted, so the code is a little huge (about 20 lines without the comments).

awk '# First record line:
     # Storing all column names into elements, including
     # the first column name
     NR == 1 {firstcol=$1;element[$1]++}

     # Each line starting with the second one are datas
     # Occurrences are counted with an indexed array
     # count[x][y] contains the count of Element y for the Gene x
     NR > 2 {element[$2]++;count[$1][$2]++} 

     # Done, time for displaying the results
     END {
       # Let us display the first line, column names
       ## Left-justify the first col, because it is text
       printf "%-10s ", firstcol
       ## Other are counts, so we right-justify
       for (i in element) if (i != firstcol) printf "%10s ", i
       printf "\n"
       
       # Now an horizontal bar
       for (i in element) {
           c = 0
       while (c++ < 10) { printf "-"}
       printf " ";
       } 
       printf "\n"

       # Now, loop through the count records
       for (i in count) {
         # Left justification for the column name
         printf "%-10s ", i ;
         for(j in element)
           # For each counted element (ie except the first one),
           # print it right-justified
           if (j in count[i]) printf "%10s", count[i][j]
         printf "\n"
       }
     }' tab-separated-input.txt

Result:

Gene            G-box        MYC 
---------- ---------- ---------- 
STBZIP10            4         3
STBZIP1             2         3

1 This solution requires GNU awk for arrays of arrays (count[$1][$2] syntax) - Thanks to Ed Morton

Upvotes: 2

Timus
Timus

Reputation: 11351

With the file in the form (named input.csv here):

    Gene     Element   
 ---------- ---------- 
  STBZIP1    G-box     
  STBZIP1    G-box     
  STBZIP1    MYC       
  STBZIP1    MYC       
  STBZIP1    MYC       
  STBZIP10   MYC       
  STBZIP10   MYC       
  STBZIP10   MYC       
  STBZIP10   G-box     
  STBZIP10   G-box     
  STBZIP10   G-box     
  STBZIP10   G-box

this

import pandas as pd

df = pd.read_csv('input.csv', delim_whitespace=True, skiprows=1)
df.columns = ['Gene', 'Element']
df['Count'] = 1
df = df.pivot_table(index='Gene', columns='Element', aggfunc=sum)
print(df)

gives you

         Count    
Element  G-box MYC
Gene              
STBZIP1      2   3
STBZIP10     4   3

Upvotes: 2

Related Questions