XfidZddlZddlZddlZddlZddlmZddlmZdZdZ dZ dZ d Z d Z dd ZGd dZGddeZGddZdS)zModule to control GenePop.N) _Argument)AbstractCommandlinec` t|S#t$rt|cYSwxYw)zDGet a float from a token, if it fails, returns the string (PRIVATE).)float ValueErrorstrtoks =lib/python3.11/site-packages/Bio/PopGen/GenePop/Controller.py _gp_floatr s=Szz 3xx --c` t|S#t$rt|cYSwxYw)zBGet a int from a token, if it fails, returns the string (PRIVATE).)intrrr s r _gp_intrs<3xx 3xxr c(|}d|vr+|dkrtd|vrdS|}d|v+d|dD}d|D}|}g}|dkrd|dD} ||d d |d d Dt |d fn?#t $r2||d dgt|zd fYnwxYw|}|dk||fS) Nz --No data)NNcg|] }|dk| Sr.0xs r z+_read_allele_freq_table..+sFFFQa2ggqggg c,g|]}t|Sr)rrs r rz+_read_allele_freq_table..,s+++awqzz+++rcg|] }|dk| Srrrs r rz+_read_allele_freq_table..0s777qqBwwwwwrrc,g|]}t|Srr rs r rz+_read_allele_freq_table..3s>>>QIaLL>>>r)readline StopIterationrstripsplitappendrrlen)flineallelestablepartss r _read_allele_freq_tabler.#s ::<>%"+>>>b @R@RS     ? ? ? LL%(TFS\\$91= > > > > > ?zz||""$$ "** E>s!AD%%9E! E!cg}|}d|vr*|}d|v*|}d|vrd|vr|dkrd|dD}g}t|D]O\}} ||||+#t $r||YLwxYw|t ||}d|vr d|vr|dk|S)N---===rcg|] }|dk| Srrrs r rz_read_table..Bs666aa2gggggrr)r#r%r& enumerater'rtuple)r)funsr,r*toksr-ir s r _read_tabler8;sr E ::<<   D t  zz||""$$ t   ::<<   D t  T 1 1dbjj664::c??666oo " "FAs " WT!WS\\**** " " " S!!!!! " U5\\"""zz||""$$ t  T 1 1dbjj Ls$C&&DDc2g}|}|dkrh|dd|dDD|}|dkh|S)Nrc,g|]}t|Srr rs r rz)_read_triangle_matrix..RsUUUy||UUUrcg|] }|dk| Srrrys r rz)_read_triangle_matrix..Rs-T-T-TAAQSGGaGGGrr)r#r%r'r&)r)matrixr*s r _read_triangle_matrixr?Ns F ::<<   D "** UU-T-TC-T-T-TUUUVVVzz||""$$ "** Mrci}|}d|vsd|vr&|}td|dDdz }t |D]}|}d|dddD}g}|D]J} |t |&#t$r|dYGwxYwt|D]\} } | ||dz| f<|S)Nr0r1cg|] }|dk| Srrrs r rz0_read_headed_triangle_matrix..\s:::!r''!'''rrr!cg|] }|dk| Srrrs r rz0_read_headed_triangle_matrix.._s:::a!r'''''r) r#r%r(r&ranger'r rr3) r)r>headernlinesline_popr*vals clean_valsvalcol_pop clean_vals r _read_headed_triangle_matrixrLWsr F ZZ\\ " "F %6//$$&& ::V\\#..::: ; ;a ?F&MM 8 8zz||""$$::4::c??122.:::  ( (C (!!)C..1111 ( ( (!!$''''' ("+J"7"7 8 8 GY.7FHqL'* + + 8 Ms+"DD0/D0Fc Z|}|rd}nd}|dkr||r|||t|tt t t t t tg}i}|D]D}t|dkr d||d<!|dd\}} } } } } | dkrd} | | | | | f||<E|S|}|dkt)NzLocus zPop : rrr"-) r#lstrip startswithr8rr rr(r$)streamis_locus has_fisherr*hookr,locientrylocuspsefis_wcfis_rhstepss r _hw_funcr^ks9 ??  D "** ;;== # #D ) )  OO    OO    OO   iIy'SVWED ? ?u::>>%)DqNN:?*7E1b&&%Syy!"#R">DKKK  ' "*** rc,eZdZdZddZdZdZdZdS) _FileIteratorzReturn an iterator which crawls over a stream of lines with a function (PRIVATE). The generator function is expected to yield a tuple, while consuming input Ncl||_|t||_n||_||_d|_dSNF)funcopenrRfnamedone)selfrcrehandles r __init__z_FileIterator.__init__s8 >u++DKK!DK  rc0|jrd|_t|SNT)rfr$rgs r __iter__z_FileIterator.__iter__s 9 DI  rc,||SN)rcrls r __next__z_FileIterator.__next__syyrcj|jtj|jdSro)rRcloseosremovererls r __del__z_FileIterator.__del__s.  $*rro)__name__ __module__ __qualname____doc__rirmrprurrr r`r`s_     rr`c&eZdZdZddZdZdZdS)_GenePopCommandlinez4Return a Command Line Wrapper for GenePop (PRIVATE).NGenepopc tdgddtdgddtdgddtd gd td gd td gdtdgdtdgdtdgdtdgdg |_tj||fi||dddS)NcommandzGenePop option to be calledT) is_requiredmodezShould always be batchinputz Input fileDememorizationzDememorization step BatchNumberzNumber of MCMC batches BatchLengthzLength of MCMC chainsHWtestszEnumeration or MCMCIsolBDstatisticzIBD statistic (a or e)MinimalDistancezMinimal IBD distanceGeographicScalez Log or Linearz Mode=Batch)r parametersrri set_parameter)rg genepop_dircmdkwargss r riz_GenePopCommandline.__init__s yk#@d S S S vh 8d K K K wi4 @ @ @ '(*? @ @ }o'? @ @ }o'> ? ? yk#8 9 9 ()+C D D ()+A B B ()? ; ;   $T399&999 6<00000rcr|dddd|DzdS)z`Set the menu option. Example set_menu([6,1]) = get all F statistics (menu 6.1) r~z MenuOptions=.c34K|]}t|VdSro)rrs r z/_GenePopCommandline.set_menu..s(0M0MAQ0M0M0M0M0M0MrN)rjoin)rg option_lists r set_menuz_GenePopCommandline.set_menusK  ~0M0M0M0M0M(M(MM     rc8|dd|zdS)zSet the input file name.rz InputFile=N)rrgres r set_inputz_GenePopCommandline.set_inputs# 7L5$899999r)Nr|)rvrwrxryrirrrrr r{r{sL>>1111    :::::rr{ceZdZdZd'dZd'dZd'dZ d(d Z d(d Z d(d Z d(d Z d)dZ d(dZ d(dZ d*dZdZ d*dZ d*dZ d*dZ d*dZdZdZdZdZdZdZdZdZd Zd+d$Zd+d%Zd+d&ZdS),GenePopControllerz5Define a class to interface with the GenePop program.Nc.t||_dS)zInitialize the controller. genepop_dir is the directory where GenePop is. The binary should be called Genepop (capital G) N)r{ controller)rgrs r rizGenePopController.__init__s.k::rcJi}||d<||d<||d<||durd|d<nd|d<|S)NrrrT EnumerationrMCMCr)rgdememorizationbatches iterations enum_testoptss r _get_optszGenePopController._get_optssP!/ %](]  D  "/Y"(Y rc h|i}tj}tj}tj||j|tj|r|j |n*|j |tj z|z|D]6}|j ||dzt||z7|tj|tj|dS)N=)rsgetcwdtempfilemkdtempchdirrrpathisabsrseprrshutilrmtree)rg extensionsoptionrercwdtemp_diropts r _run_genepopzGenePopController._run_genepops  <Dikk#%%    ((( 7==   < O % %e , , , , O % %cBFlU&: ; ; ; K KC O ) )#sSy3tCy>>/I J J J J    hrT'c|||||}||gd|g||d} t| ||zS)a;Use Hardy-Weinberg test for heterozygote deficiency/excess (PRIVATE). Returns a population iterator containing a dictionary where dictionary[locus]=(P-val, SE, Fis-WC, Fis-RH, steps). Some loci have a None if the info is not available. SE might be none (for enumerations). r!c,t|jdSrbr^rRrls r hw_funcz4GenePopController._test_pop_hz_both..hw_funcsDK// /r)rrr`) rgretypeextrrrrrrs r _test_pop_hz_bothz#GenePopController._test_pop_hz_boths`$~~ngz9MM 3%!TE4888 0 0 0Weck222rc|||||}||gd|g||d} t||z5} | } d| vr| } d| vt | t t t t g} dddn #1swxYwYt||z5} | } d| vr| } d| vt | t t t t g}dddn #1swxYwYt||z5}|} d| vr|} d| v|||||} tdd| d DD\}}}dddn #1swxYwY| ||||ffS) aUse Global Hardy-Weinberg test for heterozygote deficiency/excess (PRIVATE). Returns a triple with: - A list per population containing (pop_name, P-val, SE, switches). Some pops have a None if the info is not available. SE might be none (for enumerations). - A list per loci containing (locus_name, P-val, SE, switches). Some loci have a None if the info is not available. SE might be none (for enumerations). - Overall results (P-val, SE, switches). r!c\t|jttttgSro)r8rRrr rls r hw_pop_funcz;GenePopController._test_global_hz_both..hw_pop_func&st{S)Y ,RSS Srz by populationNzby locusz all locusc34K|]}t|VdSror rs r rz9GenePopController._test_global_hz_both..<s9$$!" ! $$$$$$rcg|] }|dk| Srrr<s r rz:GenePopController._test_global_hz_both..=s&M&M&MQQ"WWqWWWrr) rrrdr#r8rr r%r4r&)rgrerrrrrrrrf1r*pop_pf2loc_pr)rYrZswitchess r _test_global_hz_bothz&GenePopController._test_global_hz_both s,~~ngz9MM 3%!TE4888 T T T%#+   L";;==D!--{{}}"--S)Y $JKKE  L L L L L L L L L L L L L L L %#+   L";;==DD(({{}}D((S)Y $JKKE  L L L L L L L L L L L L L L L %#+   !::<H>c 8||dd||||S)*Use Hardy-Weinberg test for heterozygote deficiency. Returns a population iterator containing a dictionary where dictionary[locus]=(P-val, SE, Fis-WC, Fis-RH, steps). Some loci have a None if the info is not available. SE might be none (for enumerations). r!z.Drrgrerrrrs r test_pop_hz_deficiencyz(GenePopController.test_pop_hz_deficiencyB+%% 1dI~w    rc 8||dd||||S)rz.Errs r test_pop_hz_excessz$GenePopController.test_pop_hz_excessRrrFc|||||}||gddg||d}d} tj|dz|dzt ||dzt | |dzfS)aUse Hardy-Weinberg test based on probability. Returns 2 iterators and a final tuple: 1. Returns a loci iterator containing: - A dictionary[pop_pos]=(P-val, SE, Fis-WC, Fis-RH, steps). Some pops have a None if the info is not available. SE might be none (for enumerations). - Result of Fisher's test (Chi2, deg freedom, prob). 2. Returns a population iterator containing: - A dictionary[locus]=(P-val, SE, Fis-WC, Fis-RH, steps). Some loci have a None if the info is not available. SE might be none (for enumerations). - Result of Fisher's test (Chi2, deg freedom, prob). 3. Final tuple (Chi2, deg freedom, prob). r!c.t|jddSrkrrls r hw_prob_loci_funcz=GenePopController.test_pop_hz_prob..hw_prob_loci_funcsDKt44 4rc.t|jddS)NFTrrls r hw_prob_pop_funcz.hw_prob_pop_funcsDK55 5rz.Pz.P2)rrrcopyfiler`) rgrerrrrrrrrs r test_pop_hz_probz"GenePopController.test_pop_hz_probbs4~~ngz9MM 3%!Q555 5 5 5 6 6 6  eem444 +UT\ : : *EEM : :  rc 8||dd||||S)a Use Global Hardy-Weinberg test for heterozygote deficiency. Returns a triple with: - An list per population containing (pop_name, P-val, SE, switches). Some pops have a None if the info is not available. SE might be none (for enumerations). - An list per loci containing (locus_name, P-val, SE, switches). Some loci have a None if the info is not available. SE might be none (for enumerations). - Overall results (P-val, SE, switches). rNz.DGrrs r test_global_hz_deficiencyz+GenePopController.test_global_hz_deficiency+(( 1eY   rc 8||dd||||S)aUse Global Hardy-Weinberg test for heterozygote excess. Returns a triple with: - A list per population containing (pop_name, P-val, SE, switches). Some pops have a None if the info is not available. SE might be none (for enumerations). - A list per loci containing (locus_name, P-val, SE, switches). Some loci have a None if the info is not available. SE might be none (for enumerations). - Overall results (P-val, SE, switches) z.EGrrs r test_global_hz_excessz'GenePopController.test_global_hz_excessrrc8||||}|dgddg||d}d}t|dz}|} d| vr|} d| vt j|dz|dzt|dz} | } d| vr| } d| vd| vr| } d| vt ||dz|t ||dz| fS) zHTest for linkage disequilibrium on each pair of loci in each population.z.DISrr!cd}|j}|dkrd|_td|dD}|d|d|d}}}t |ds||}}d }||kr ||kr|dz }|d d kr||||fdfSt|d t|d t|d } } } ||||f| | | ffS)NrTcg|] }|dk| Srrrs r rzBGenePopController.test_ld..ld_pop_func..:::!!r''A'''rrrr!r start_locus1r"rNorNr) rRr#r%rfr$r&hasattrr r) rg current_popr*r6poplocus1locus2r start_locus2rYrZrs r ld_pop_funcz.GenePopController.test_ld..ld_pop_funcsK;''))0022Drzz  ##::tzz#:::D"&q'47DGC400 !-3Vl  %%&L*@*@q Aw$"C&&)94??'Q00)DG2D2DgdSTgFVFV8rAff%52x7HH Hrc|j}|dkrd|_td|dD}|d|d}} t |dt|dt |d }}}n#t$r ||fdfcYSwxYw||f|||ffS) NrTcg|] }|dk| Srrrs r rz>GenePopController.test_ld..ld_func..rrrrrrrNr) rRr#r%rfr$r&r rr)rgr*r6rrchi2dfrYs r ld_funcz*GenePopController.test_ld..ld_funcs;''))0022Drzz  ##::tzz#:::D!!Wd1gFF .'Q00'$q'2B2BIdSTgDVDV!b . . .'---- .F#dB]2 2s0?B00CC----z.DI2z Locus pair)rrrdr#rrr`) rgrerrrrrrrr*rs r test_ldzGenePopController.test_ldsQ~~ngzBB 6(QFE4888 I I I$ 3 3 3%&. ! !{{}}D  ;;==DD  777 %&. ! !{{}}$&&;;==D$&&D  ;;==DD   +uv~r : : '56>2 6 6  rct)z4Provision for creating Genotypic contingency tables.NotImplementedErrorrs r create_contingency_tablesz+GenePopController.create_contingency_tables!!rct)z8Provision for Genic differentiation for all populations.rrgrerrrs r test_genic_diff_allz%GenePopController.test_genic_diff_all "!rct)z=Provision for Genic differentiation for all population pairs.rrs r test_genic_diff_pairz&GenePopController.test_genic_diff_pairrrct)z>A}#,QWWQZZ#8#8  PRVWWA}$-aggajj$9$9! +T22A}!!''!**-- +T22A}!!''!**-- +T22A}!!''!**-- /66A} )!''!** 5 5  %&.!!!!2E5%VVsA  AAc|dgddg|d}d}tj|dz|dzt||dz}t||dz}||fS)aCalculate allele and genotype frequencies per locus and per sample. Parameters: - fname - file name Returns tuple with 2 elements: - Population iterator with - population name - Locus dictionary with key = locus name and content tuple as Genotype List with (Allele1, Allele2, observed, expected) (expected homozygotes, observed hm, expected heterozygotes, observed ht) Allele frequency/Fis dictionary with allele as key and (count, frequency, Fis Weir & Cockerham) - Totals as a pair - count - Fis Weir & Cockerham, - Fis Robertson & Hill - Locus iterator with - Locus name - allele list - Population list with a triple - population name - list of allele frequencies in the same order as allele list above - number of genes Will create a file called fname.INF INFrr!c t|dr |j}|`n|j}i}|dkr4|}d|vr |j|fSt jd|}||d}|d}t|ds||_ t|dr!|j |kr|j}||_|` |`||fS||_n|j}g}|j}d |vrd |vr|j}d |v|d krt jd |}|| t|dt|dt|d t|dfn|j}|j}|d kd|vr|j}d|vt|dd} |j}t|dd} |j}t|dd} |j}t|dd} |j}d|vr|j}d|v|j}i} d}d|vrd| dD}|ddkr?t|dt|dt|d f}nTt|dt|dt|d f| t|d<|j}d|v|| | | | f| |f||<|dk4d|_t)Nold_linerz,Tables of allelic frequencies for each locusz.*Pop: (.+) Locus: (.+)r!r first_locuscurr_poprzGenotypes Obs. z% +([0-9]+) , ([0-9]+) *([0-9]+) *(.+)rrNzExpected number of ho&z Sample countrcg|] }|dk| Srrrs r rzTGenePopController.calc_allele_genotype_freqs..pop_parser..KKK!177A777rrrTotT)rrrRr#r%rr matchr rr'rr r&rfr$)rgr* loci_contentr rrXold_pop geno_listm2expHoobsHoexpHeobsHefreq_fis overall_fisrGs r pop_parserz@GenePopController.calc_allele_genotype_freqs..pop_parserfs|tZ(( .}MM{++--L"**{{}}ATII=,66!:DAA$++a..//11C!KKNNE"4771+0(tZ009+u44&*mG,0DM $ 0 $ #*L#88$'DMM;//11D {++--$$'t33;//11D(t33dll"I4PPB~!(( ' 4 4 ' 4 4 ' 4 4 )"((1++ 6 6  ${3355 ;//11Ddll .T99;//11D.T99!$rss),,{++--RSS **{++--!$rss),,{++--RSS **{++--$D00;//11D%D00{++--" D((KKt{{}}':':3'?'?KKKDAw%''#DG,,%d1g..%d1g..' $DG,,%d1g..%d1g..6a!1!12  ;//11DD(( E5%0 ' U#["**fDI rcR|j}|dkr{|}tjd|}|1|d}t |j\}}|||fS|j}|dk{d|_tNrz Locus: (.+)r!T) rRr#r%r r r r.rfr$)rgr*r rXr+r,s r locus_parserzBGenePopController.calc_allele_genotype_freqs..locus_parsers;''))D"**{{}}66$!KKNNE%.fis_func..s<00"! 000000rcg|] }|dk| Srrr<s r rzMGenePopController._calc_diversities_fis..fis_func..s%""""#AQSGGAGGGrrT) rRr#r%r r r r8rr r4r&rfr$)rgr*r rX fis_table avg_qinteravg_fiss r fis_funcz9GenePopController._calc_diversities_fis..fis_funcss;''))D"**{{}}ImT22=GGAJJEK((***$ (<(<(>(>>>$d{*K((*** + c9i%K!!IK((****/00""'+{';';'='='C'CC'H'H"""000++'J !)Z@@{++--)"***DI r) rrdr#r%rQr8rr r`)rgrerr)r*r8 avg_Qintrar9s r _calc_diversities_fisz'GenePopController._calc_diversities_fiss3 3%!Q/// %#+   $!::<.rrrNrrc|t|dr |j}|`n|j}d}d}d}d}d}d}|dkr|}|dr@|||_||||||fS|dd}n=|dr)t|dd}n|dr)t|dd}n|d r)t|dd}n|d r)t|dd}nE|d r0t|dd}||||||fS|j}|dk|||||||fS|j d |_ t) N last_linerz Locus::r!zFis^=rzFst^=zFit^=z 1-Qintra^=z 1-Qinter^=T) rrDrRr#r%rQr&rPr rrrfr$)rgr*rXfisfstfitqintraqinters r procz,GenePopController.calc_fst_all..proc-s*t[)) .~NN{++--ECCCFF"**{{}}??:..@()-$c3VVCC $ 3 2 9 9 ; ;__W-- @#DJJsOOA$677CC__W--@#DJJsOOA$677CC__W--@#DJJsOOA$677CC__\22@&tzz#q'9::FF__\22@&tzz#q'9::F #sC??{++--'"**( c3VV;; K     DI r) rrdr#rQr%r&r rr`) rgrer)r*r6allFisallFstallFitrKs r calc_fst_allzGenePopController.calc_fst_all s 6(QFE222 %&. ! ! $Q::<s r loci_funcz2GenePopController.calc_fst_pair..loci_funcas;''))D"**{{}}Ind33=GGAJJE9$+FFF &=({++--"**DI r) rrdr#r%rQrLrsrtr`)rgrer)r*avg_fstrSs r calc_fst_pairzGenePopController.calc_fst_pairVs" 66*QFE::: %&. ! ! $Q::<:1==Gzz|| "** $ $ $ $ $ $ $ $ $ $ $ $ $ $ $  %&.!!!Y77@@sA-B''B+.B+ct)zPProvision for estimating spatial structure from Allele size for all populations.rrs r calc_rho_allzGenePopController.calc_rho_allrrrct)zUProvision for estimating spatial structure from Allele size for all population pairs.rrs r calc_rho_pairzGenePopController.calc_rho_pairwrraLogh㈵>c |gdd|g||||dt|dz5}||||t|}||t|}|t jd|} t| d} t| d} ||t jd |} t| d} t jd |} t| d} t| d}d d d n #1swxYwYtj |d ztj |d ztj |dz||| | f| | |ffS)z5Calculate isolation by distance statistics (PRIVATE).).GRArQ.ISOrA)rrr)rr_za = (.+), b = (.+)r!rz b=(.+)z.*\[(.+) ; (.+)\]NrQr^) rrdr#r?r r r%r r rsrt)rgresubstatscalemin_distr)estimatedistancer rZbbbbblowbbhighs r _calc_ibdzGenePopController._calc_ibd{sv  $ $ $ H #+#(#'  %&. ! ! /Q JJLLL JJLLL JJLLL JJLLL,Q//H JJLLL JJLLL,Q//H JJLLLH11::<<3F3F3H3HIIE%++a..))A%++a..))A JJLLL JJLLLHY (;(;(=(=>>E5;;q>>**BH3QZZ\\5H5H5J5JKKEekk!nn--Eu{{1~~..F' / / / / / / / / / / / / / / /( %&.!!! %&.!!! %&.!!!Aq6Bv+>>>sH(I((I,/I,c4||d|||S)zCalculate isolation by distance statistics for diploid data. See _calc_ibd for parameter details. Note that each pop can only have a single individual and the individual name has to be the sample coordinates. rrjrgrerarbrcs r calc_ibd_diploz GenePopController.calc_ibd_diplo~~eQeX>>>rc4||d|||S)zCalculate isolation by distance statistics for haploid data. See _calc_ibd for parameter details. Note that each pop can only have a single individual and the individual name has to be the sample coordinates. rArlrms r calc_ibd_haploz GenePopController.calc_ibd_haplororro)Trrr)Frrr)rrr)rZr[r\) rvrwrxryrirrrrrrrrrrrrrrrrr1r;r=r?rOrUrWrYrjrnrqrrr rrs??;;;;        ,3333>2/2/2/2/lSW     "SW     (( ( ( ( XSW    *SW    (2 2 2 2 j""" CG""""CG""""CG""""CG""""&W&W&WR_&_&_&B'I'I'IT999 """ JMJMJMZAAA8""" """#?#?#?#?L??????????rr)F)ryrsr rrBio.Applicationrrr rr.r8r?rLr^r`r{rrrr rssT! %%%%%%//////0&(<@:::::-:::Bj ?j ?j ?j ?j ?j ?j ?j ?j ?j ?r