class Lexique

module lexique

This class is used for all of the computations of partitions, partitionings and HMM analysis.

A lexique is a set of descriptors. The prediction functions, used for all the computations, are defined by the descriptors.

Beware: The descriptors #33 and #94 should not be used (see descriptors).

Optionaly, some values can be put for the transitions between the descriptors: these values are used for HMM computations.

Construction

__init__

Optional keywords:

str=s

builds from string s, using read_str;

alpha=a

uses the letters of string a, in use with option str (see input);

Lprop=l

builds from Lproportion l, using read_Lprop;

fic=f

builds from filename f, using read_nf;

fprop=f

builds from filename f of Lproportion, using read_Lprop;

read_nf

builds from a filename and optionaly a string of letters (see input);

read_str

builds from a string and optionaly another string of letters (see input).

read_Lprop

builds from a Lproportion, each Proportion being translated as in read_prop, and the transitions between descriptors are valued by the logarithms of the transitions proportions.

Handling

is_empty

returns True if there is no descriptor, False otherwise;

__len__

returns the number of descriptors;

ls_num

returns the list of the numbers of the descriptors;

met_au_net

removes repetitions in patterns of descriptors (like ACAC cleaned in AC) and repetitions of descriptors (like A A cleaned in A);

__iter__

iterates over the descriptors, using for d in lx;

__delitem__

removes the Descripteur of given number;

__getitem__

gets the Descripteur of given number by COPYING it;

init_trans

initialises the array of the transition probabilities;

g_trans

gets the logarithm of the transition probability between Descripteurs of given numbers;

__iadd__

add the copies of the Descripteur of a given Lexique. If the number of a Descripteur is already in the Lexique, this Descripteur is not added;

__setitem__

• either sets the Descripteur of given number to given Descripteur by COPYING it (if the number is already used in the Lexique, the corresponding descriptor is replaced by the new one);
• or builds a descriptor pattern given a tuple of numbers (numbers that must not have been alreadyy used in the Lexique), and a tuple or a list of Descripteurs.

For example:

>>> import lexique >>> l=lexique.Lexique(str="1:A 2,3:BC") >>> print l 3,2:CB 1:A >>> import descripteur >>> d=descripteur.Descripteur(3,str="Z") >>> print d Z >>> l[2]=d >>> print l 3,2:CZ 1:A >>> d.read_str("P") >>> print l 3,2:CZ 1:A >>> e=descripteur.Descripteur(3,str="Y") >>> l[5]=e 5:Y >>> print l 5:Y 3,2:CZ 1:A >>> l[4,3]=e,d Bad descriptor number 3 already used >>> l[4,6]=e,d 4,6:YP >>> print l 6,4:PY 5:Y 3,2:CZ 1:A >>> del l[3] >>> del l[2] >>> print l 6,4:PY 5:Y 1:A >>> l[3,2]="X",e 3,2:XY >>> print l 2,3:YX 6,4:PY 5:Y 1:A #########################

prediction

computes the maximum prediction on a data, without using between-descriptors transitions;

Optional keywords:

deb=d

sets the first position of the segment on which the prediction is computed (default: 0);

fin=f

sets the last position of the segment on which the prediction is computed (default: last position-1);

If deb>fin, it returns 0.

val_max

computes the maximum sum of predictions on a data, without using between-descriptors transitions; in that case, the maximal prediction is computed on each position, and the sum of these values is returned;

Optional keywords:

deb=d

sets the first position of the segment on which the prediction is computed (default: 0);

fin=f

sets the last position of the segment on which the prediction is computed (default: last position-1);

If deb>fin, it returns 0.

llh

computes the best log-likelihood on a Sequence.

Optional keywords:

deb=d

sets the first position of the segment on which the log-likelihood is computed (default: 0);

fin=f

sets the last position of the segment on which the log-likelihood is computed (default: last position-1);

If deb>fin, it returns 0.

ls_evalue

returns the dictionnary of tuple of (the pattern of) numbers of descriptors, prediction of this (pattern of) descriptor on the data, for all descriptors, on a given data.

Optional keywords:

deb=d

sets the first position of the segment on which the predictions are computed (default: 0);

fin=f

sets the last position of the segment on which the predictions are computed (default: last position -1 );

If deb>fin, it returns 0.

windows

computes the list of the best predictions on the data in a sliding window of a given size and with steps of a given length;

log_likelihood

computes the list of the log-likelihoods of the data given the numbers of segment, up to a given number of segments.

Input-Output

• Output

  Specific format is:

  description numbers_of_descriptors_separated_by_commas:       outputs_of_the_descriptors_in_the_pattern_separated by spaces.   optional lines of transitions costs between the descriptors, in format: number_of_descriptor,number_of_descriptor     cost

  examples   1,2:+{A(-1)CG}T 3:C     1:A 2:C 3:G 4:T     1,2,3:A|`C(0.3)GT(0.2)A'{C(0.5)A(-1.5)} 5:+(CG) 10:`T(-0.23)!(0.1)'     1:A 2:T   1,1 -2.3 1,2 -4.1 2,2 7.2 2,1 0

  __str__

  outputs in specific format.

• Input

  Input is in the same format as the output, plus:

  • there is no need to give numbers to the descriptors. In that case, they will be automatically numbered;
  • character $ is a wildcard, and cannot be used as a standard letter. It is used in method read_str, and is set successively on all of the letters of the second string argument of this method. For example:
    • lit_str("$", "ACG") builds Lexique: A C G
    • lit_str("A$A", "TG") builds Lexique: ATA AGA
    • lit_str("`A$$'", "CGT") builds Lexique:
      `ACC' `ACG' `ACT' `AGC' `AGG' `AGT' `ATC' `ATG' `ATT'