[aldino]

Yes, still interested!

Was just about to start doing this myself but I thought I'd check this forum again...actually, would really like to use BFAST for solid but just don't have the computing resources available to me...

anyway, thanks again...

[ulz_peter]

Yes. Works nicely with GATK (at least in my case).

[KevinLam]

Nice! Have u tried it with gatk?

[ulz_peter]

Just in case anyone's still interested:
I wrote a small (rather dirty) python script to integrate those tags. It is slow and relies on the fact that bwa solid2fastq, bwa aln and bwa samse step doesn't change the order of the reads. Anyway it might be of interest to someone...

#! /usr/bin/python

#ADD CS and CQ tags from original CSfasta and csqual file

import sys

#try getting file names from comand line
try:
  SAMfile = sys.argv[1]
  csfastafile = sys.argv[2]
  qualfile = sys.argv[3]
  outputfile=sys.argv[4]
except: 
  print ("Usage: ./add_CSCQ.py <input SAM> <input csfasta> <input csqual> <output SAM>")
  sys.exit()

#try open files specified in command line
try:
   SAM = open(SAMfile)
   csfasta = open (csfastafile)
   qual = open (qualfile)
   output = open (outputfile, "w")
except:
  print ("Couldn't open SAMfile")
  sys.exit()

#reading the first lines of the three files
SAMline = SAM.readline()
cs = csfasta.readline()

#iterate till no comment
startcs=cs[0:1]
while startcs=='#':
  cs=csfasta.readline()
  startcs=cs[0:1]
  
cq = qual.readline()
startcq=cq[0:1]
while startcq=='#':
  cq=qual.readline()
  startcq=cq[0:1]

count = 0

#iterate through all the files and add CS / CQ tags in the reads
#assuming solid2fastq didn't change the order of the reads
while SAMline:
  info=SAMline.split()
  start=SAMline[0:1]
  alignment=SAMline[:-1]
  
  if (((count % 100000) == 0) and (count != 0)):
    print count, "alignments processed"

#print out header section
  if start == '@':    
    output.write(alignment+'\n')

#print alignment section and add CS and CQ tags
  else:
    #read csfasta file until no comments
    cs=csfasta.readline()
    cq=qual.readline()
    cs="\tCS:Z:"+cs
    cs=cs[:-1]
    
    #encode quals to sanger quals
    cq=cq[:-1]
    intquals=cq.split()
    asciiquals=""
    for quality in intquals:
       quality = int (quality)
       ascii=chr(quality+33)
       asciiquals=asciiquals+ascii
    cq="\tCQ:Z:"+asciiquals
    
    #write alignment to file
    output.write(alignment+cs+cq+'\n')
    cs=csfasta.readline()
    cq=qual.readline()
  SAMline = SAM.readline()
  count = count + 1
  
SAM.close()
csfasta.close()
qual.close()

[drio]

bfast+bwa only replaces the match step. Postprocess is the same as in traditional bfast. You will have those tags.

[Todd Scheetz]

Thanks Kevin! I will take a look.

BTW, did you ever try the BWA alignment from within BFAST (bfast+bwa)? That would seem to solve the problem -- assuming it includes the CS and CQ tags. I will be trying that soon.

Todd

[KevinLam]

Ah found it!

http://kevin-gattaca.blogspot.com/2010/07/shiny-new-tool-to-index-ngs-reads-g-sqz.html

[KevinLam]

Hi Todd,
Unfortunately, I decided to switch mapper for SOLID reads in the end.
I am trying to find the fastq indexer program that I intended to use for this with scripts to post process the bam
but I can only find this http://ivory.idyll.org/blog/mar-10/s...ving-sequences

Hope it helps!

[Todd Scheetz]

Solution?

Hi Kevin,

Did you work out a program to integrate the color space data into the SAM files for BWA alignments? If so, could you share? I am running into the same issue, and would like avoid re-implementation if possible.

Thanks,
Todd

[nilshomer]

Thanks Nils,
I have an example of CS CQ tags

VAB_S1332068_1358_1351 131 1 227 255 25M = 1373 1171 CTAACCCCTAACCCTAACCCTAAAC !A@B?@@@CAC?@?AAC?
??B@AA! RG:Z:TG133 CS:Z:G3230100023010023010023001 CQ:Z:<<<<<<<<<<;:<;* MD:Z:25 OQ:Z:!@@@@@@@@@@@@@@@@@@@@@@@!

So am I correct in saying that CS and CQ are essentially the original csfasta and qual line?
or at least bfast outputs it in this way?

additionally
but if I were to do the same I might have problems as bwa does trimming?

They should be the qualities from the csfasta/qual lines. I don't think it has to be matched to the trimming. When it means "original", it means unaltered "original" values IMHO.

[KevinLam]

Thanks Nils,
I have an example of CS CQ tags

VAB_S1332068_1358_1351 131 1 227 255 25M = 1373 1171 CTAACCCCTAACCCTAACCCTAAAC !A@B?@@@CAC?@?AAC?
??B@AA! RG:Z:TG133 CS:Z:G3230100023010023010023001 CQ:Z:<<<<<<<<<<;:<;* MD:Z:25 OQ:Z:!@@@@@@@@@@@@@@@@@@@@@@@!

So am I correct in saying that CS and CQ are essentially the original csfasta and qual line?
or at least bfast outputs it in this way?

additionally
but if I were to do the same I might have problems as bwa does trimming?

[nilshomer]

I see...
So far, I only have these info. so presumably I have to reverse the order of the original CS and CQ if the read is mapped in another direction?

the trimming bit might indeed be a problem though even if trying to construct a query db of the original csfasta and qual files isn't computationally intensive.


Color read sequence on the same strand as the reference 4
CS Z
Color read quality on the same strand as the reference; encoded in the same way as <QUAL> 4
CQ Z

On a raw SOLiD read, the first nucleotide is the primer base and the first color is the one between the primer base
and the first nucleotide from the sample being sequenced. The primer base and the first color must be present in CS.

I don't ever match the direction of the CS/CQ tags to the reference, since the reverse (not compliment) is not symmetric. The adapter would then be the last base.

[KevinLam]

I see...
So far, I only have these info. so presumably I have to reverse the order of the original CS and CQ if the read is mapped in another direction?

the trimming bit might indeed be a problem though even if trying to construct a query db of the original csfasta and qual files isn't computationally intensive.


Color read sequence on the same strand as the reference 4
CS Z
Color read quality on the same strand as the reference; encoded in the same way as <QUAL> 4
CQ Z

On a raw SOLiD read, the first nucleotide is the primer base and the first color is the one between the primer base
and the first nucleotide from the sample being sequenced. The primer base and the first color must be present in CS.

[nilshomer]

Hi Nils, can you share how bfast generates the CS tag?

I think I probably will have to write a post alignment script to add this tag

The CSFASTQs I use as input are not "double-encoded" but instead contain the origin color sequence (unadulterated) and color qualities. This allows the CS/CQ tags to filled out easily. BWA "double-encodes" the color sequence and does some trimming too, thus making it impossible to recover the CS/CQ without going back to the original data (CSFASTA and QUAL).

[KevinLam]

It doesn't. Contact the developer if you want this feature in the future.

Hi Nils, can you share how bfast generates the CS tag?

I think I probably will have to write a post alignment script to add this tag