How to introduce a spacer in a transcription factor PWM matrix?

Question

[Gema Sanz]

Hi,

I need to modify the PWM matrix of a certain transcription factor to allow a spacer of different length, i.e., from 1 to 20 nucleotides, in a matrix like this for example:

A  [13006 75198     0     0  4556     0 74715  8654 60151 99494     0 ]

C  [10026  5868     0     0     0 99494  5478 51954 39343     0 36038 ]

G  [33617 18428     0     0 75531     0 10015     0     0     0  2043 ]

T  [42845     0 99494 99494 19407     0  9286 38886     0     0 61413 ]

But I'm not sure about how modify the concensus PWM from example from JASPAR to include the nucleotides, I don't know how to specify the nucleotide frequency or probability... Any ideas?

Thank you very much in advance.

Gema

Answer 1

[Asaf]

You can't represent a gapped motif using a PWM. An HMM would fit better. Practically you can download the sequences used for this motif generation from JASPAR and predict a gapped motif using GLAM2 from the MEME suite.

[Gema Sanz]

Thanks for your answer! I will take a look

[Gema Sanz]

I checked your suggestion but I think is not exactly what I want to do. The approach that I want to follow is described in one paper as follows:

The position weight matrix (PWM) from the Jaspar database19 described the p53 binding motif, composed by two decameric half sites. In order to account for the possibility of a spacer, an array of 15 PWMs was designed, containing a string of 1–15 uniformly distributed nucleotides between the two halves.

But they don't describe how to construct the matrices.

[Asaf]

They generated 15 matrices if I understood correctly. The gap between the two parts would contain background nucleotide distribution. What is your goal eventually?

[Gema Sanz]

To do the same, but creating 18 instead of 15.