keep read association with mpileup

Question

[gernophil]

Hey,

is there a way to make all bases that are from the same read in one vertical line using mpileup? So making this

chr2    96  C   5   .,.,,
chr2    97  A   4   .,.,
chr2    98  C   6   .,.,,,
chr2    99  C   8   .,.,,,..
chr2    00  A   9   .,,,..,,.
chr2    01  C   5   ..,,.
chr2    02  C   5   ,.,,,
chr2    03  T   4   .,,,

to this (made up example)

chr2    96  C   5     .,.,,  
chr2    97  A   4     .,. ,  
chr2    98  C   6   .,.,, ,  
chr2    99  C   8   .,.,,,.. 
chr2    00  A   9   .,,,..,,.
chr2    01  C   5   ..,    ,.
chr2    02  C   5   ,.,    ,,
chr2    03  T   4   .,,    ,

To make it look a little bit like in IGV? I know this would be impossible for large regions, but I am just looking at very tiny region up to 20 or max 30 bp.

EDIT: To make it more clear, let's assume this is our reference sequence:

CGATGCTAGC

And these are our NGS reads:

CGATG
GATGCT
AXGCX
GCTAG
TAGC

These should be mapped like this

CGATGCTAGC

CGATG
 GATGCT
  AXGCX
    GCTAG
      TAGC

and default mpileup would look like this:

1   C   .
2   G   ..
3   A   ...
4   T   ..X
5   G   ....
6   C   ...
7   T   .X..
8   A   ..
9   G   ..
10  C   .

Here you cannot see that the both mismatches (X) are from the same read, so that's why I want to make it look like this:

1   C   .
2   G   ..
3   A   ...
4   T   ..X
5   G   ....
6   C    ...
7   T    .X..
8   A      ..
9   G      ..
10  C       .

Here you can see that the both mismatches are from the same read, because they are in the same possition in the mpileup string.

[Ram]

Can you explain the changes you've made in your made up example? I cannot understand your requirement.

[gernophil]

I’ll try :). I’ll introduced some spaces so that the bases that belong to the same read are always in the same position in the mpileup string, if a read does not contain this position this would be space. I want to get an overview, of the general sequencing quality in one position so if there’s a rare mutation I can check, if the whole read is bad or, if this is really likely a mutation with very low AF. I’m on my mobile right now and I will try to make a figure to illustrate this better.

[gernophil]

Ok, I added some text and I hope that makes it more clear :).

[Ram]

Thank you, your requirement is clear now. I don't know much in this domain, but I have a feeling that you may have to write a custom script, which is not a big deal. You'll merely need to create a matrix from the read-based pileup, replace matches with dots and mismatches with X/,/whatever, then transpose the matrix. I'll try and create an R example.

EDIT: The script is more challenging that I thought it would be, especially given the fact that I start with the alignment you show in your example and not your actual starting point. Please wait for others to weigh in, they may have better options for you.