ó ¯V^c@sÆddlZddlZddlZddlZddlZddlZdZdZedkrhdGHne e d„Z d„Z d„Z d „Zd „Zd „Zd „Zd „Ze d„ZdS(iÿÿÿÿNsGraham Alvare, Brian FristenskysSun Mar 18 14:57:59 CDT 2012t__main__sPHYLIP Library modulecCs"t|dƒ}tddƒ}|j|jƒƒ|jƒtddƒ}|j|jƒƒ|jƒ|rtddƒ}|j|jƒƒ|jƒ|r|jdƒ|jdƒ|jddƒ|jd dƒ|jd dƒ|jd dƒ|jd dƒqn|jƒd S(s Merge messages from MSGFILE to the outfile generated by a Phylip program. If extra is included as a parameter, this appends a message indicating that the output is a consensus tree, whose branch lengths are the bootstrap values, NOT the actual branch lengths. twtMSGFILEtrtoutfilesoutfile.consensets s<>>>> THIS TREEFILE IS A CONSENSUS TREE, WHOSE BRANCH LENGTHSs5>>>> ARE BOOTSTRAP VALUES, NOT ACTUAL BRANCH LENGTHS.s>>>> TO GENERATE BRANCH LENGTHSs6>>>> USE TREE FILE AS INPUT FOR DNAML OR OTHER PROGRAMs>>>> USING THE USERTREE OPTIONNs s (topent writelinest readlinestclosetwrite(t outfile_pathtconsensetextrat h_OUTFILEt h_MSGFILEth_smalloutfileth_outfile_consense((s'/home/psgendb/BIRCHDEV/script/phylip.pyt merge_msgs(     cCs3t|dƒ}|jƒjƒd}|jƒ|S(sIRead first line of a Phylip file to find out how many sequences there areRi(RtreadlinetsplitR (tinfileth_infiletNUMSEQ((s'/home/psgendb/BIRCHDEV/script/phylip.pyt get_numseq:s cCsjt|ƒ}|dkrE||krE|jdƒ|j|dƒnd}|jdt|ƒdƒ|S(s-Make sure OUTGROUP is not greater than NUMSEQitos s OUTGROUP = so (tintR tstr(tOUTGROUPRtcomfilet h_msgfilet tempoutgroup((s'/home/psgendb/BIRCHDEV/script/phylip.pyt do_outgroupBs  cCs.tjddƒ}t|ddƒd}|S(s1Generate a random integer as needed by Phylip programs. These numbers are used by Phylip programs as seeds for a random number stream. They must be odd, in the form 4n + 1. Return value is an integer between 0 and 2e16 -1 (which is 65535). Although the Phylip Main document claims that 32-bit random numbers are acceptible, at least one program, PARS, will only take 16-bit random numbers. Use floor division by 4 to generate a whole number quotient, multiply the quotient by 4 to generate an even number, and add 1 to make it odd. iiÿÿii(trandomtrandintR(tpseedtprand((s'/home/psgendb/BIRCHDEV/script/phylip.pyt phylip_randomNs c Csgtjƒ} dttjƒƒ} tj| ƒtj|tjj| dƒƒtj | ƒt ddƒ} t |ƒdksŽt |ƒdkr—d}n| j dƒ| j t|ƒdƒt |ƒd ksßt |ƒdkrèd }nt |ƒd kr!| j d ƒ| j t|ƒdƒn|d kr=| j d ƒn—|d krf| j d ƒ| j d ƒnn|dkr| j d ƒ| j d ƒnE|dkrÅ| j d ƒ| j d ƒ| j d ƒn|dkrÔn|dkrÿ|dkrG| j dƒqGnH|dkr7| j dƒ|dkrG| j dƒqGn|dkrf| j dƒ| j dƒd}ná|dkr¢| j dƒ| j dƒ| j dƒd}n¥|dkrë| j dƒ| j dƒ| j dƒ| j dƒd}n\|dkrA| j dƒ| j dƒ| j dƒ| j dƒ| j dƒd}nd}| j dƒ| j t|ƒdƒ|dkr‡| j dƒn| j dƒt ƒ} | j t| ƒdƒ| jƒt dd ƒ} tjdgd| ƒ}|jƒ| jƒ|dkr'tjd tjj| | ƒƒntjd!tjj| | ƒƒtj | ƒtj| tƒd"S(#sRun Seqboot. Notes: SEQBOOT reads interleaved sequences by default, but can read sequential files using the "I" setting. By default, SEQBOOT writes datasets to outfile, but will write weights to 'outweights' if you set the "S" option. Weight files are always sequential. sSEQBOOT.RtSeqbootComfileRiids% s isB tmsd RtRtgtstbtyesss tdsj tpstnotpotpwtrewsr si sy tseqboottstdint outweightsRN(tostgetcwdRtgetpidtmkdirtshutiltcopytpathtjointchdirRRR R%R t subprocesstPopentwaittmovetrmtreetTrue(tINFILEtDATATYPEtRSEEDtMETHODt REPLICATEStPERCENTt BLOCKSIZEt OUTWEIGHTSt OUTFORMATtOUTFILEtSTARTDIRtTEMPDIRt comfile_httempseedtp((s'/home/psgendb/BIRCHDEV/script/phylip.pyR3_s˜   $  $                                              " cCskttƒƒ}|jdt|ƒd|dƒ|jdƒ|j|dƒ|jt|ƒdƒdS(s´Jumble - When multiple datasets are analyzed, protpars automatically jumbles, and prompts for a random number seed for jumbling. Otherwise, jumbling must be explicitly set.sJUMBLING SEQUENCE ORDER s ITERATIONS, SEED=s tjNsj (RR%R (RRtNUMJUMt tempjseed((s'/home/psgendb/BIRCHDEV/script/phylip.pytjumbleàs # c CsÏtƒ}t|ƒ}t|ƒ} |dkr=|jdƒnŽ|dkr"|dkrŸ|jd| d|dƒt|ƒdkr¼|jd |dƒq¼n|jd | d|dƒt|ƒd krê|jd |d dƒntjddƒtd||||||d|dƒ n©|dkrË|dkrZ|jd| d|dƒnI|dkr†|jd| d|dƒn|jd| d|dƒtd||||||d|dƒ ndS(säRun SEQBOOT to generate resampled datasets. Output is weights, used by Phylip programs to generate datasets on the fly. If you want SEQBOOT to generate actual datasets instead of weights, use weightless_resample instead.tnt s R+R-sRESAMPLING: Bootstrap, s REPLICATES, SEED=isResampling in blocks of s$RESAMPLING: Delete-half Jacknifing, idsPartial Resampling: spercent of sites sampleds infile.tempRR,tweightsR.R0R1s0RESAMPLING: Permute species for each character, s%RESAMPLING: Permute character order, s$RESAMPLING: Permute within species, R/Ns (R+R-(spsspospw(R%RR RR:tcopyfileR3( RHRJRIRKRRFRMtBseedtBseedStrtRepStr((s'/home/psgendb/BIRCHDEV/script/phylip.pyt stdresampleìs,      (     c Csptƒ}t|ƒ}t|ƒ} |dkr=|jdƒn/|dkrl|dkru|jd | d |dƒnÏ|dkr¡|jd | d |dƒn£|dkrÍ|jd | d |dƒnw|dkrù|jd | d |dƒn|jd| d |dƒt|ƒdkrD|jd|ddƒntd||||||d|dƒ ndS(s€ Generate terminal input for running seqboot to generate randomized sequence files, rather than weight files and run seqbootRXRYs R+R-R.R0R1s0RESAMPLING: Permute species for each character, s REPLICATES, SEED=s%RESAMPLING: Permute character order, s$RESAMPLING: Permute within species, sRESAMPLING: Bootstrap, s$RESAMPLING: Delete-half Jacknifing, idsPartial Resampling: spercent of sites sampleds infile.tempR/RNs (R+R-spsspospw(R%RR RR3( RHRJRIRKt msgfile_hRFRMR\R]R^((s'/home/psgendb/BIRCHDEV/script/phylip.pytweightless_resamples$             cCsí|r'ttjj|dƒdƒ}n tj}t|dƒ}|jƒ}|jƒd}|djdƒdkrÁx0|D](}|jdƒdkr{|d}q{q{W|j t |ƒd ƒnx|D]}|j |ƒqÈW|jƒd S( s·Used by dnaml.py and protml.py. Make sure that treefile begins with number of trees on first line of file. If first line in file has parentheses, the number must be added.tintreetaRit(iÿÿÿÿt;is N( RR6R<R=tsyststdoutRR tfindR R(tUFNRPRtinfile_htufn_ht ufn_linestsc_counttline((s'/home/psgendb/BIRCHDEV/script/phylip.pytufn4s!     (R6tos.pathR!R:R?Rft __author__t__date__t__name__RDtFalseRRR R%R3RWR_RaRo(((s'/home/psgendb/BIRCHDEV/script/phylip.pyts$       %   (