Does preg_split allow mismatches?

JoeZar

I want to split a DNA sequence (an random string of A,C,G and T)I want to split a DNA sequence (a long random string formed by 4 letters - A, C, G or T-, which may be very long) by using a small sequence (with same 4 letters).

Example:
Sequence to split ($seq):
$seq="ACTAGTCCGCTTGCGCATC..."

Length of sequence my be >5 milion

$pattern="/CGCTAGTCAG/"
Length of typical pattern:10-30

I will split it this way:
$an_array=preg_split ($pattern,$seq)

That is OK, but my next step is to allow mismatches.

How? With preg_split? with a different command?

paulnaj

Need a bit more detail about what 'mismatches' are.

JoeZar

Mismach means error in pattern recognition (one mismatch, one error, two mismatches, tewo errors etc).

For example:
sequence: AAAAATCGTAGAAAAA
pattern:TCGAAG.

the pattern will recognize the sequence only if one mismacth is allowed.

AAAAATCGTAGAAAAA
TCGAAG

paulnaj

One way of doing it ...

This will allow a mismatch in the 1st character ...

$pattern="/[ACGT]GCTAGTCAG/"

This will allow a mismatch in the 2nd character ...

$pattern="/C[ACGT]CTAGTCAG/"

You could program this for every char ... but it's a lot of work for the server.

JoeZar

This is the way I have found to split the sequence by using the subsequence TCGAA, and allowing one mismatch:

$pattern="/(.CGAA|T.GAA|TC.AA|TCG.A|TCGA.)/";
$an_array=preg_split($pattern,$string);

If possible, I would like to find a way to allow more mismacthes(2, 3 etc) by using a more effective system.

paulnaj

I'm going to have to have a think about this.

What's your priority? Is it speed/efficiency or flexibility/ease of use.

JoeZar

My priority is speed/efficiency

paulnaj

My first thought would be to pass over the DNA-string many times.

The first pass would look exclusively for 'full' matches, the 2nd would look for the '1 mismatches', etc.

On each pass, replace the '(mis)match' with a placeholder (single character) to keep the integrity of the remaining portions.

However, the order in which you check matches will have a bearing on the result.

For example, suppose you have a DNA-string 'ACGTGATC', and you want to match two different strings 'CG' and 'GT' against it.

That is, suppose you check for 'CG' first ...
'CG' matches 'ACGTGATC' leaving 'AXTGATC'
'GT' now doesn't match.

But, if you check for 'GT' first ...
'GT' matches 'ACGTGATC' leaving 'ACXGATC'
'GT' now doesn't match.

I'd be really interested in trying to help you do this, but I keep coming up with questions. It might be easier to do this by email.