XfU.6dZddlmZdZGddZdS)aCode to parse BIG GenePop files. The difference between this class and the standard Bio.PopGen.GenePop.Record class is that this one does not read the whole file to memory. It provides an iterator interface, slower but consuming much mess memory. Should be used with big files (Thousands of markers and individuals). See http://wbiomed.curtin.edu.au/genepop/ , the format is documented here: http://wbiomed.curtin.edu.au/genepop/help_input.html . Classes: - FileRecord Holds GenePop data. Functions: ) get_indivc$t|}|S)zfParse a file containing a GenePop file. fname is a file name that contains a GenePop record. ) FileRecord)fnamerecords =lib/python3.11/site-packages/Bio/PopGen/GenePop/FileParser.pyreadr s   F McZeZdZdZdZdZdZdZdZdZ dZ d Z d Z d Z d Zd ZdS)ra`Hold information from a GenePop record. Attributes: - marker_len The marker length (2 or 3 digit code per allele). - comment_line Comment line. - loci_list List of loci names. Methods: - get_individual Returns the next individual of the current population. - skip_population Skips the current population. skip_population skips the individuals of the current population, returns True if there are more populations. get_individual returns an individual of the current population (or None if the list ended). Each individual is a pair composed by individual name and a list of alleles (2 per marker or 1 for haploid data). Examples:: ('Ind1', [(1,2), (3,3), (200,201)] ('Ind2', [(2,None), (3,3), (None,None)] ('Other1', [(1,1), (4,3), (200,200)] cXd|_g|_||_|dS)zInitialize the class.N) comment_line loci_listr start_read)selfrs r__init__zFileRecord.__init__As.  r czd}|jdzg}|d|jdz|j}|j}|jd||dd}|r| }|dur|dn|durd}n|\}}|||d|D]} |d| D]g} | d } t| } t| |kr*d d | g} t| |k*|| h|d|| ||d |S) zReturn (reconstructs) a GenePop textual representation. This might take a lot of memory. Marker length will be 3.  rzPop TF, N0r ) rappendjoinr current_pop current_ind_handleseek skip_headerget_individualstrlen seek_position) r marker_lenreprrmoreresnamemarkersmarkeralaStrs r__str__zFileRecord.__str__Hs   4'( 499T^,,t3444& &  !  7 !%%''Cd{{ 7#### # g 4    3%))FJJsOOO$)):!$B"2ww!$ii*44#%77C;#7#7D"$ii*44 4(((( ) 4   ' !( ; 444wws||r cRt|j|_|j|_|jdd}|d}|j ||jD]J}|}| dkrn*|j |Ktdd|_ d|_dS)z/Start parsing a file containing a GenePop file.rr rPOPz;No population data found, file probably not GenePop relatedrN)openrrreadlinerstriprreplacesplitrextendupperr ValueErrorrr)rsample_loci_lineall_locilines rrzFileRecord.start_readns DJ''  L1133::<< <002299;;CCCLL#))#.. h'''L  D;;==Dzz||u$$ N ! !$ ' ' ' 'M r cd|_d|_|jD]/}|dkrdS0dS)z,Skip the Header. To be done after a re-open.rr/N)rrrr2r6rr:s rrzFileRecord.skip_headersXL  D{{}}""$$--.  r c|jd||dkr||dz}|dk|dkr!||dz}|dkdSdS)zSeek a certain position in the file. Arguments: - pop - pop position (0 is first) - indiv - individual in pop rN)rrrskip_populationr )rpopindivs rr#zFileRecord.seek_positions ! Agg  " " " 1HCAggaii    ! ! ! QJEaiiiiiir c|jD]Q}|dkrdS|}|dkr|xjdz c_d|_dSRdS)zBSkip the current population. Returns true if there is another pop.r Fr/r>rTN)rr2r6rrr<s rr?zFileRecord.skip_populationsuL  Drzzuu;;==Dzz||u$$  A%  #$ tt%   r c|jD]p}|}|dkr|xjdz c_d|_dS|xjdz c_t |\}}}||fcSdS)a\Get the next individual. Returns individual information if there are more individuals in the current population. Returns True if there are no more individuals in the current population, but there are more populations. Next read will be of the following pop. Returns False if at end of file. r/r>rTF)rr2r6rrr)rr: indiv_name allele_listignores rr zFileRecord.get_individualsL / /D;;==Dzz||u$$  A%  #$ tt  A%  2;D/// K!;....ur cXt|j}t|d5}||jdz|jD]}||dzd}|}d}|r||kr||dz }n|dur|dz }d}n|r|dd}|\} } || dz| D]} |d | D]g} | d } t| } t| d kr*d d | g} t| d k*|| h|d|}|d d d d S#1swxYwYd S)zRemove a population (by position). Arguments: - pos - position - fname - file to be created with population removed wrrTr>POP FrrNrrr ) r rr0writerrr r?r!r"r)rposrold_recflocuscurr_popl_parser start_popr(r)r*r+r,s rremove_populationzFileRecord.remove_populations$tz"" %   4 GGD%, - - - * & & %%%%H--//HI 4s??++---MHH4'' A $( $.GGG,,,(-I(0 gs +++&-..FGGCLLL&,..#%:),B'*2ww&)$ii!mm+-77C;+?+?D'*$ii!mm ! . "11333 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4sE-FF#&F#cPt|j}t|d5}||jdz|j}||=|D]}||dz|}|d|r|dur|dn|\}} ||dzd} | D]} | |kr| dz } | dz } |d| D]g} | d } t| } t| d kr*d d | g} t| d k*|| h|d|}|d d d d S#1swxYwYd S) zRemove a locus by position. Arguments: - pos - position - fname - file to be created with locus removed rHrrITrrr>rNrrr ) r rr0rJrrr r!r"r)rrKrrLrMrrNrPr(r) marker_posr*r+r,s rremove_locus_by_positionz#FileRecord.remove_locus_by_positionstz"" %   4 GGD%, - - -)I#" & & %%%%--//H GGG    4t##GGG$$$$$,MD'GGD3J'''!"J") * *%,,&!OJ$"a  "(**B!z%(#&r77D"%d))a--')wwT{';';#&d))a--GGDMMMM *GGDMMM"1133- 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4sE)FF"Fct|j}t|d5}||jdz|j}||t}|D]}||=| ||D]}||dz| } |d| r| dur|dn| \} } || dzd} | D]} | |vr| dz } | dz } |d| D]g}|d }t|}t|d kr*d d |g}t|d k*||h|d| } | d d d d S#1swxYwYd S) zRemove a set of loci by position. Arguments: - positions - positions - fname - file to be created with locus removed rHrrITrrr>rNrrr )r rr0rJrrsortreversesetaddr r!r"r)r positionsrrLrMrposSetrKrNrPr(r)rTr*r+r,s rremove_loci_by_positionz"FileRecord.remove_loci_by_positionshtz"" %  # 4 GGD%, - - -)I NN         UUF   cN 3" & & %%%%--//H GGG    4t##GGG$$$$$,MD'GGD3J'''!"J") * *%//&!OJ$"a  "(**B!z%(#&r77D"%d))a--')wwT{';';#&d))a--GGDMMMM *GGDMMM"1133- 4# 4# 4# 4# 4# 4# 4# 4# 4# 4# 4# 4# 4# 4# 4# 4# 4# 4# 4sF7G))G-0G-cxt|jD]$\}}||kr|||dS%dS)zRemove a locus by name. Arguments: - name - name - fname - file to be created with locus removed N) enumeraterrU)rr(rirNs rremove_locus_by_namezFileRecord.remove_locus_by_nameAsV"$.11  HAu}}--a777  r cg}t|jD]\}}||vr|||||dS)zRemove a loci list (by name). Arguments: - names - names - fname - file to be created with loci removed N)r_rrr])rnamesrr[r`rNs rremove_loci_by_namezFileRecord.remove_loci_by_namePs` !$.11 $ $HAu~~  ### $$Y66666r N)__name__ __module__ __qualname____doc__rr-rrr#r?r rRrUr]rardr rrr&s4$$$L."   ,)4)4)4V'4'4'4R,4,4,4\    7 7 7 7 7r rN)rhBio.PopGen.GenePoprr rrir rrksn$)(((((v7v7v7v7v7v7v7v7v7v7r