Mask x number of bases WITHIN sequence prior to alignment

[SEQond]

Message moved to correct section
http://seqanswers.com/forums/showthread.php?t=22014

Hi all,

As you may see from the picture I have this QC from all R2 reads of my Paired End sequenced samples. Due to a technical error during the sequencing I am ending up with 30+ R2 reads with serious errors in the middle of the sequence.

Do you know any way to mask (or to allow mismatch at) a specific number of bases (2-3) at a specific position WITHIN the fragment length prior to alignment? Biostrings is an option that I would prefer not to use for reasons of speed.

Can what you propose be selectively applied to only one of the two reads in the paired end samples?

It would be ideal if this could be directly applied directly with Bowtie like the trimming left/right that already exists as an inherent option.

[HESmith]

ELAND should be able to mask these bases during alignment, since the initial error-free segment is longer than the seed. Include USE_BASES Y41n3Y*n in the config file to mask bases 42-44 (plus the terminal base, which is the default). The one caveat is if the error results from fluidics/chemistry issues, then the phasing after the error may be incorrect.

If this doesn't work or you prefer an alternative aligner, you could convert the reads into pseudo-paired end data. Use bases 1-41 as read one, then use bases 45-100 as read two. Filter the aligned data on expected criteria (i.e., both reads map to same chromosome and orientation, position of read 2 = read 1 + 44 [or + 41-44 if phasing is off]).

[SEQond]

ELAND should be able to mask these bases during alignment, since the initial error-free segment is longer than the seed. Include USE_BASES Y41n3Y*n in the config file to mask bases 42-44 (plus the terminal base, which is the default). The one caveat is if the error results from fluidics/chemistry issues, then the phasing after the error may be incorrect.

If this doesn't work or you prefer an alternative aligner, you could convert the reads into pseudo-paired end data. Use bases 1-41 as read one, then use bases 45-100 as read two. Filter the aligned data on expected criteria (i.e., both reads map to same chromosome and orientation, position of read 2 = read 1 + 44 [or + 41-44 if phasing is off]).

Can ELAND in this way align Paired End sequences while at the same time masking selectively bases of only a one of the two reads?

To be honest I would prefer a BW based aligner (Bowtie,BWA, and SOAP2)
Thanks for your answer

[SEQond]

Possibly bowtie 2 is the answer to the issue

also look here or here