o Rŀgi@sdZddlZddlZddlZddlZddlmZddlmZddZddZ d d Z d d Z d dZ ddZ dddZGdddZGdddeZGdddZdS)zModule to control GenePop.N) _Argument)AbstractCommandlinecC&zt|WStyt|YSw)zDGet a float from a token, if it fails, returns the string (PRIVATE).)float ValueErrorstrtokr Q/var/www/html/myenv/lib/python3.10/site-packages/Bio/PopGen/GenePop/Controller.py _gp_float    r cCr)zBGet a int from a token, if it fails, returns the string (PRIVATE).)intrrrr r r _gp_intr rc Cs|}d|vr|dkrtd|vrdS|}d|vsdd|dD}dd|D}|}g}|dkrd d|dD}z||d d d|d d Dt|d fWntyx||d dgt|d fYnw|}|dks=||fS)Nz --No data)NNcSg|]}|dkr|qSrr .0xr r r +z+_read_allele_freq_table.. cSg|]}t|qSr )rrr r r r,cSrrr rr r r r0rrcSrr r rr r r r3r)readline StopIterationrstripsplitappendrrlen)flineallelestablepartsr r r _read_allele_freq_table#s0 & "  r*c Csg}|}d|vr|}d|vs |}d|vrod|vro|dkrodd|dD}g}t|D]\}}z ||||Wq8tyU||Yq8w|t||}d|vrod|vro|dks(|S)N---===rcSrrr rr r r rBrz_read_table..r)rr!r" enumerater#rtuple)r%funsr(r&toksr)ir r r r _read_table;s&      r2cCsTg}|}|dkr(|dddd|dDD|}|dks |S)NrcSrr rrr r r rRrz)_read_triangle_matrix..cSrrr ryr r r rRrr)rr!r#r")r%matrixr&r r r _read_triangle_matrixNs $ r6c Csi}|}d|vsd|vr|}tdd|dDd}t|D]E}|}dd|dddD}g}|D]}z |t|WqBty[|dYqBwt|D] \} } | ||d| f<q`q(|S)Nr+r,cSrrr rr r r r\rz0_read_headed_triangle_matrix..rrcSrrr rr r r r_r) rr!r$r"ranger#r rr-) r%r5headernlinesline_popr&vals clean_valsvalcol_pop clean_valr r r _read_headed_triangle_matrixWs&     r@Fc Cs|}|r d}nd}|dkrg||r_|||t|tttttttg}i}|D]*}t|dkrAd||d<q2|dd\}} } } } } | dkrSd} | | | | | f||<q2|S|}|dkst)NzLocus zPop : rrr-) rlstrip startswithr2rr rr$r )streamis_locus has_fisherr&hookr(locientrylocuspsefis_wcfis_rhstepsr r r _hw_funcks0 rQc@s2eZdZdZd ddZddZddZd d ZdS) _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 NcCs0||_|dur t||_n||_||_d|_dSNF)funcopenrEfnamedone)selfrTrVhandler r r __init__s   z_FileIterator.__init__cCs|jrd|_t|SNT)rWr rXr r r __iter__sz_FileIterator.__iter__cCs ||SN)rTr\r r r __next__s z_FileIterator.__next__cCs|jt|jdSr^)rEcloseosremoverVr\r r r __del__s z_FileIterator.__del__r^)__name__ __module__ __qualname____doc__rZr]r_rcr r r r rRs    rRc@s*eZdZdZd ddZddZdd ZdS) _GenePopCommandlinez4Return a Command Line Wrapper for GenePop (PRIVATE).NGenepopc Kstdgdddtdgdddtdgdddtd 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 parametersrrZ set_parameter)rX genepop_dircmdkwargsr r r rZs        z_GenePopCommandline.__init__cCs$|ddddd|DdS)z`Set the menu option. Example set_menu([6,1]) = get all F statistics (menu 6.1) rjz MenuOptions=.cs|]}t|VqdSr^)rrr r r sz/_GenePopCommandline.set_menu..N)rvjoin)rX option_listr r r set_menusz_GenePopCommandline.set_menucCs|dd|dS)zSet the input file name.rmz InputFile=N)rvrXrVr r r set_inputsz_GenePopCommandline.set_input)Nri)rdrerfrgrZrrr r r r rhs   rhc@s<eZdZdZdCddZdCddZdCddZ dDd dZ dDddZ dDddZ dDddZ  dEddZ dDddZ dDddZ dFddZddZ dFd d!Z dFd"d#Z dFd$d%Z dFd&d'Zd(d)Zd*d+Zd,d-Zd.d/Zd0d1Zd2d3Zd4d5Zd6d7Zd8d9ZdGd=d>ZdGd?d@ZdGdAdBZdS)HGenePopControllerz5Define a class to interface with the GenePop program.NcCst||_dS)zInitialize the controller. genepop_dir is the directory where GenePop is. The binary should be called Genepop (capital G) N)rh controller)rXrwr r r rZszGenePopController.__init__cCsDi}||d<||d<||d<|dur |durd|d<|Sd|d<|S)NrnrorpT EnumerationrqMCMCr )rXdememorizationbatches iterations enum_testoptsr r r _get_optsszGenePopController._get_optscCs|duri}t}t}t||j|tj|r&|j |n |j |tj ||D]}|j ||dt ||q3|t|t |dS)N=)ragetcwdtempfilemkdtempchdirrrpathisabsrseprvrshutilrmtree)rX extensionsoptionrVrcwdtemp_diroptr r r _run_genepops     zGenePopController._run_genepopT'c Cs<|||||}||gd|g||dd} t| ||S)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). rcSs t|jdSrSrQrEr\r r r hw_funcs z4GenePopController._test_pop_hz_both..hw_func)rrrR) rXrVtypeextrrrrrrr r r _test_pop_hz_bothsz#GenePopController._test_pop_hz_bothcCs|||||}||gd|g||dd} t||!} | } d| vr.| } d| vs&t| ttttg} Wdn1sAwYt||!} | } d| vr]| } d| vsUt| ttttg}Wdn1spwYt||B}|} d| vr|} d| vs|||||} tdd d d | d DD\}}}Wdn1swY| ||||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). rcSst|jttttgSr^)r2rErr r\r r r hw_pop_func&sz;GenePopController._test_global_hz_both..hw_pop_funcz by populationNzby locusz all locuscsr{r^rrr r r r|<s z9GenePopController._test_global_hz_both..cSrrr r3r r r r=rz:GenePopController._test_global_hz_both..r) rrrUrr2rr r!r.r")rXrVrrrrrrrrf1r&pop_pf2loc_pr%rLrMswitchesr r r _test_global_hz_both s@  z&GenePopController._test_global_hz_bothc C||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). rz.DrrXrVrrrrr r r test_pop_hz_deficiencyB z(GenePopController.test_pop_hz_deficiencyc Cr)rz.Errr r r test_pop_hz_excessRrz$GenePopController.test_pop_hz_excessFc Csf|||||}||gddg||dd}dd} t|d|dt||dt| |dfS) 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). rcSst|jddSr[rr\r r r hw_prob_loci_funcz=GenePopController.test_pop_hz_prob..hw_prob_loci_funccSst|jddS)NFTrr\r r r hw_prob_pop_funcrz.hw_prob_pop_funcz.Pz.P2)rrrcopyfilerR) rXrVrrrrrrrrr r r test_pop_hz_probbs  z"GenePopController.test_pop_hz_probc Cr)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). rAz.DGrrr r r test_global_hz_deficiencyz+GenePopController.test_global_hz_deficiencyc Cr)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.EGrrr r r test_global_hz_excessrz'GenePopController.test_global_hz_excessc Cs||||}|dgddg||dd}dd}t|d}|} d| vr0|} d| vs(t|d|d t|d } | } d | vrP| } d | vsHd| vr\| } d| vsTt||d|t||d | fS) zHTest for linkage disequilibrium on each pair of loci in each population.z.DISrrc Ssd}|j}|dkrd|_tdd|dD}|d|d|d}}}t|d s6||}}d }||krB||krB|d7}|d d krP||||fdfSt|d t|d t|d} } } ||||f| | | ffS)NrTcSrrr rr r r rrzBGenePopController.test_ld..ld_pop_func..rrrr start_locus1rrNorAr) rErr!rWr r"hasattrr r) rX current_popr&r0poplocus1locus2r start_locus2rLrMrr r r ld_pop_funcs    (z.GenePopController.test_ld..ld_pop_funccSs|j}|dkrd|_tdd|dD}|d|d}}zt|dt|d t|d }}}WntyH||fdfYSw||f|||ffS) NrTcSrrr rr r r rrz>GenePopController.test_ld..ld_func..rrrrrAr) rErr!rWr r"r rr)rXr&r0rrchi2dfrLr r r ld_funcs, z*GenePopController.test_ld..ld_func----z.DI2z Locus pair)rrrUrrrrR) rXrVrrrrrrrr&rr r r test_lds*  zGenePopController.test_ldcCt)z4Provision for creating Genotypic contingency tables.NotImplementedErrorrr r r create_contingency_tablesz+GenePopController.create_contingency_tablescCr)z8Provision for Genic differentiation for all populations.rrXrVrrrr r r test_genic_diff_allz%GenePopController.test_genic_diff_allcCr)z=Provision for Genic differentiation for all population pairs.rrr r r test_genic_diff_pairrz&GenePopController.test_genic_diff_paircCr)z.pop_parser..rrTotT)rrrErr!rrmatchrrr#rr r"rWr )rXr& loci_contentrrrKold_pop geno_listm2expHoobsHoexpHeobsHefreq_fis overall_fisr;r r r pop_parserfs                                     Sz@GenePopController.calc_allele_genotype_freqs..pop_parsercSsl|j}|dkr1|}td|}|dur(|d}t|j\}}|||fS|j}|dks d|_tNrz Locus: (.+)rT) rErr!rrrr*rWr )rXr&rrKr'r(r r r locus_parsers     zBGenePopController.calc_allele_genotype_freqs..locus_parserz.INFz.IN2)rrrrR)rXrVrrpop_iter locus_iterr r r calc_allele_genotype_freqs4s#] z,GenePopController.calc_allele_genotype_freqscCs||gddg|t||1}|}|dkr:|}|dr2t|ttttg}t|ttg}|}|dksWdn1sDwYdd}t|||||fS)NrrrzGStatistics per sample over all loci with at least two individuals typedcSs|j}|dkre|}td|}|dur\|d}|jd|jvr,|dfS|jt|jttttg}|jt dddd|j d DD\}}||||fS|j}|dks d |_ t ) Nrz Locus: (.+)rz No completecsr{r^rrr r r r|s  zLGenePopController._calc_diversities_fis..fis_func..cSrrr r3r r r rszMGenePopController._calc_diversities_fis..fis_func..rT) rErr!rrrr2rr r.r"rWr )rXr&rrK fis_table avg_qinteravg_fisr r r fis_funcs0        z9GenePopController._calc_diversities_fis..fis_func) rrUrr!rDr2rr rR)rXrVrr%r&r avg_Qintrarr r r _calc_diversities_fiss  z'GenePopController._calc_diversities_fiscCs ||dS)z/Compute identity-base Gene diversities and Fis.z.DIV)rrr r r "calc_diversities_fis_with_identitys z4GenePopController.calc_diversities_fis_with_identitycCr)zAProvision to Computer Allele size-based Gene diversities and Fis.rrr r r calc_diversities_fis_with_sizerz0GenePopController.calc_diversities_fis_with_sizec Cs|dgddg|t|de}|}|dkrn|drfdd|dD}zt|d}Wn ty=d }Ynwzt|d }Wn tyQd }Ynwzt|d }Wn tyed }Ynw|}|dksWd n1sxwYd d }|||ft||dfS)aExecute GenePop and gets Fst/Fis/Fit (all populations). Parameters: - fname - file name Returns: - (multiLocusFis, multiLocusFst, multiLocus Fit), - Iterator of tuples (Locus name, Fis, Fst, Fit, Qintra, Qinter) Will create a file called ``fname.FST``. This does not return the genotype frequencies. z.FSTrrz All:cSrrr rr r r rrz2GenePopController.calc_fst_all..rNrrcSsft|dr |j}|`n|j}d}d}d}d}d}d}|dkr|}|drB|dur8||_||||||fS|dd}nR|drQt|dd}nC|dr`t|dd}n4|d rot|dd}n%|d r~t|dd}n|d rt|dd}||||||fS|j}|dks |dur||||||fS|j d |_ t ) N last_linerz Locus::rzFis^=rzFst^=zFit^=z 1-Qintra^=z 1-Qinter^=T) rrrErr!rDr"rCr r`rWr )rXr&rKfisfstfitqintraqinterr r r proc-sF          z,GenePopController.calc_fst_all..proc) rrUrrDr!r"r rrR) rXrVr%r&r0allFisallFstallFitr r r r calc_fst_all s6    &zGenePopController.calc_fst_allcCs|ddgddg|t|d%}|}|dkr/|}|dr't|}|}|dksWdn1s9wYdd }t|dt||d|fS) zIEstimate spatial structure from Allele identity for all population pairs.z.ST2.MIGrrrzEstimates for all lociNcSsf|j}|dkr.|}td|}|dur%|d}t|j}||fS|j}|dks d|_tr) rErr!rrrr@rWr )rXr&rrKr5r r r loci_funcas     z2GenePopController.calc_fst_pair..loci_func) rrUrr!rDr@rarbrR)rXrVr%r&avg_fstrr r r calc_fst_pairVs  zGenePopController.calc_fst_paircCr)zPProvision for estimating spatial structure from Allele size for all populations.rrr r r calc_rho_allrrzGenePopController.calc_rho_allcCr)zUProvision for estimating spatial structure from Allele size for all population pairs.rrr r r calc_rho_pairwrzGenePopController.calc_rho_pairaLogh㈵>cCsh|jgdd|g||||ddt|du}||||t|}||t|}|td|} t| d} t| d} ||td |} t| d} td |} t| d} t| d}Wd n1swYt |d t |d t |d||| | f| | |ffS)z5Calculate isolation by distance statistics (PRIVATE).).GRAr.ISOr)rsrtrr)rrza = (.+), b = (.+)rrz b=(.+)z.*\[(.+) ; (.+)\]Nrr) rrUrr6rrr!r rrarb)rXrVsubstatscalemin_distr%estimatedistancerrbbbbblowbbhighr r r _calc_ibd{sD zGenePopController._calc_ibdcC||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. rr'rXrVrrr r r r calc_ibd_diploz GenePopController.calc_ibd_diplocCr()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. rr)r*r r r calc_ibd_haplor,z GenePopController.calc_ibd_haplor^)Trrr)Frrr)rrr)rrr) rdrerfrgrZrrrrrrrrrrrrrrrrrrrrrrrrr'r+r-r r r r rsf    6   ,  5    )"*M  & r)F)rgrarrrBio.Applicationrrr rr*r2r6r@rQrRrhrr r r r s"    !