0
$\begingroup$

what do curly braces in sequence motifs stand for? e.g. in

RTCRYBN{4}ACG

what is N{4}? moreover, i notice that in TRANSFAC matrix notation the N{4} is completely omitted:

NA  Abf1p
XX
DE  RTCRYBN{4}ACG
XX
P0  A     C     G     T
01  0.500 0.000 0.500 0.000 R
02  0.000 0.000 0.000 1.000 T
03  0.000 1.000 0.000 0.000 C
04  0.500 0.000 0.500 0.000 R
05  0.000 0.500 0.000 0.500 Y
06  0.000 0.333 0.333 0.333 B
XX
P1  A     C     G     T
01  1.000 0.000 0.000 0.000 A
02  0.000 1.000 0.000 0.000 C
03  0.000 0.000 1.000 0.000 G

moreover the MEME suite's transfac2meme completely ignores the second chunk of the matrix after the P1 row.

$\endgroup$
  • 3
    $\begingroup$ I'm voting to close this question as off-topic because this is about the computational aspect of Bioinformatics. While questions on interdisciplinary subjects like bioinformatics are welcome, they must focus on the biological part of the subject. $\endgroup$ – AMR Jan 15 '16 at 19:12
  • $\begingroup$ @AMR I would not consider this as off-topic because it is about understanding data from a bioinformatics repository. It is not about programming or any other software issues. $\endgroup$ – WYSIWYG Jan 16 '16 at 7:02
  • $\begingroup$ @WYSIWYG It is not asking about the biology in any way. You could just as easily be looking at astronomical data. If they had asked how can I determine if a DNA binding protein will bind to this sequence or how can I determine if this is an enhancer site, then that is a biology question. This is purely from the technical side. Also mdperry's answer answers it, so there is no point in keeping it open anyway. It also shows no research effort, because all that is needed is to Google bioinformatics nucleotide codes and the first entry gives the answer. It could be closed for several reasons. $\endgroup$ – AMR Jan 16 '16 at 17:00
  • $\begingroup$ @AMR This question is about analysis of bioinformatics data, which is on topic (akin to bioinfo data formats).. I have worked in bioinformatics a bit and I can vouch that it is not some kind of an obvious thing.. Experienced people may look at it and know, but not everyone... AFAIK we made it clear that this may not be not one of the close reasons.. $\endgroup$ – WYSIWYG Jan 16 '16 at 18:53
  • $\begingroup$ @AMR your comments are unfair -- i have tried to answer this question myself. your googling suggestion does not answer my question as i already knew what 'N' stands for but not the {4} bit. $\endgroup$ – stas g Jan 17 '16 at 23:53
4
$\begingroup$

N is the IUPAC code for any nucleotide, so in DNA sequence an N signifies any one of the four bases could be in that position.

The {4} means 4 of the previous character in the pattern, or NNNN. In Perl regular expressions \d{4} means match 4 digits in a row, so the notation is quite similar.

$\endgroup$
  • $\begingroup$ thank you for your answer. i knew what N stood for but i couldn't find what curly braces meant. i guessed that could mean NNNN but then it doesn't explain why the TRANSFAC matrix for the motif ignores these nucleotides. moreover there are motives like TNNCGTNNNNNNTGAT in the same document. why isn't the same notation used here? i.e. TNNCGTN{6}TGAT? $\endgroup$ – stas g Jan 17 '16 at 23:55
  • $\begingroup$ As to why some people prefer using curly braces vs. expanding the characters and printing them all, I have no idea (or why some programs or databases prefer one or the other). I have not used TRANSFAC very much at all; as I recall it requires a licence(?). I think what is important to keep in mind is that using a consensus sequence (the IUPAC codes) is generally not going to be as sensitive as using a Position Frequency Matrix (PFM) or a Position Weight Matrix (PWM). That matrix you show is not a real PFM/PWM--it is a PFM derived from the consensus sequence motif. $\endgroup$ – mdperry Jan 19 '16 at 3:04
  • $\begingroup$ it's a TRANSFAC formatted file in the Yeastract database which is open to public; and it just seems that they are being inconsistent with regards to their notation in the same document then. okay, i didn't realise i was using something that was't a real PWM. when you say 'less sensitive' do you mean using it for binding site predictions won't be as accurate as using the real PWM? $\endgroup$ – stas g Jan 19 '16 at 16:38
  • $\begingroup$ Let's say that there are between 10 and 100 known Abf1 binding sites in the Yeast genome. By lining them up (aligning them to each other, and summing up the number of each nucleotide at each position, and then dividing that number by the number of sequences in the alignment) we get the PFM. A simple way to create a PWM (circa 1990) is to multiply the value at each field in the matrix by the log(base 2) of the number, (m x log2(m)). (there are probably much more sophisticated transformations available now). Now you have a way of calculating the bits of information at each site in the matrix $\endgroup$ – mdperry Jan 19 '16 at 23:45
  • $\begingroup$ With a PWM you can use a program to scan along an entire chromosome, giving each site a score. If you do this (make a PFM and a PWM) do you get the matrix from TRANSFAC? If you do, great, but if you don't then you might well get more 'hits' with the real PWM. Keep in mind that without experimental confirmation of some sort none of these predicted binding sites are necessarily real binding sites, they are just sites that score above some arbitrary threshold. $\endgroup$ – mdperry Jan 19 '16 at 23:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.