Position-weight matrix
This module can be used to create position-weight matrices (PWM) to describe a DNA motif. Such matrices can then be searched on a DNA sequence, given a threshold for alignment score.
Type to represent gap-free alignments. First dimension is the sequence position, second dimension is for the alphabet. Only DNA alphabet (A, C, G, T) is supported to rows should be of length exactly four.
val flat_background : unit -> backgroundUniform distribution over A, C, G, T
val background_of_sequence : string -> float -> backgroundbackground_of_sequence seq pc estimates the base frequency in seq using pc as pseudo-counts. Typical value for pc is 0.1.
val random_background : unit -> backgroundval length : t -> intval make : count_matrix -> background -> tBuilds a PWM from a count_matrix and a background
val random : len:int -> background -> tval tandem :
?orientation:[ `direct | `inverted | `everted ] ->
spacer:int ->
count_matrix ->
count_matrix ->
background ->
ttandem orientation spacer cm1 cm2 bg builds a PWM by constructing a composite motif: it builds mat1 the PWM from cm1 under background bg (resp. mat2 from cm2 under bg), then concatenates mat1 and mat2 with spacer non scoring columns in between
val scan : t -> string -> float -> (int * float) listscan mat seq tol returns the list of positions (with corresponding scores) such that the alignment score of mat is superior to tol
val opt_scan : t -> string -> float -> (int * float) listIdentical to scan but implementing a small optimization (which is actually worse in native code but a lot better in bytecode)
val fast_scan : t -> string -> float -> (int * float) listIdentical to scan but directly implemented in C
val opt_fast_scan : t -> string -> float -> (int * float) listimproved version of fast_scan
val best_hit : t -> string -> int * floatbest_hit mat seq returns the position and score of the best alignment found in seq for the motif mat. Raise Invalid_arg if seq is shorter than mat