:>"^@sdZddlmZddlmZmZmZGdddeZGdddeZGdd d eZ Gd d d eZ Gd d d eZ GdddeZ GdddeZ GdddeZGdddeZGdddeZGdddeZGdddeZGdddeZGdddeZGd d!d!eZGd"d#d#eZGd$d%d%eZGd&d'd'eZGd(d)d)eZGd*d+d+eZGd,d-d-eZGd.d/d/eZGd0d1d1eZGd2d3d3eZGd4d5d5eZGd6d7d7eZ Gd8d9d9eZ!e"d:krdd;l#m$Z$e$d<S)=zCode to interact with and run various EMBOSS programs. These classes follow the AbstractCommandline interfaces for running programs. )print_function)_Option_SwitchAbstractCommandlinec@s%eZdZdZdddZdS)_EmbossMinimalCommandLineaBase Commandline object for EMBOSS wrappers (PRIVATE). This is provided for subclassing, it deals with shared options common to all the EMBOSS tools: Attributes: - auto Turn off prompts - stdout Write standard output - filter Read standard input, write standard output - options Prompt for standard and additional values - debug Write debug output to program.dbg - verbose Report some/full command line options - help Report command line options. More information on associated and general qualifiers can be found with -help -verbose - warning Report warnings - error Report errors - fatal Report fatal errors - die Report dying program messages Nc Ks|dk sttddgdtddgdtddgd td d gd td dgdtddgdtddgdtddgdtddgdtddgdg }y||j|_Wntk r||_YnXtj|||dS)Nz-autoZautozTurn off prompts. Automatic mode disables prompting, so we recommend you set this argument all the time when calling an EMBOSS tool from Biopython. z-stdoutstdoutzWrite standard output.z-filterfilterz+Read standard input, write standard output.z-optionsoptionszPrompt for standard and additional values. If you are calling an EMBOSS tool from within Biopython, we DO NOT recommend using this option. z-debugdebugz"Write debug output to program.dbg.z-verboseverbosez%Report some/full command line optionsz-helphelpzReport command line options. More information on associated and general qualifiers can be found with -help -verbose z-warningwarningzReport warnings.z-errorerrorzReport errors.z-dieZdiezReport dying program messages.)AssertionErrorr parametersAttributeErrorr__init__)selfcmdkwargsextra_parametersr>> cline = Primer3Commandline(sequence="mysequence.fas", auto=True, hybridprobe=True) >>> cline.explainflag = True >>> cline.osizeopt=20 >>> cline.psizeopt=200 >>> cline.outfile = "myresults.out" >>> cline.bogusparameter = 1967 # Invalid parameter Traceback (most recent call last): ... ValueError: Option name bogusparameter was not found. >>> print(cline) eprimer3 -auto -outfile=myresults.out -sequence=mysequence.fas -hybridprobe=True -psizeopt=200 -osizeopt=20 -explainflag=True The equivalent for anyone still using an older version of EMBOSS would be: >>> cline = Primer3Commandline(sequence="mysequence.fas", auto=True, hybridprobe=True) >>> cline.explainflag = True >>> cline.oligosize=20 # Old EMBOSS, instead of osizeopt >>> cline.productosize=200 # Old EMBOSS, instead of psizeopt >>> cline.outfile = "myresults.out" >>> print(cline) eprimer3 -auto -outfile=myresults.out -sequence=mysequence.fas -hybridprobe=True -productosize=200 -oligosize=20 -explainflag=True Zeprimer3cJKs6tddgdddtddgdtd d gd td d gdtddgdtddgdtddgdtddgdtddgdtddgd td!d"gd#td$d%gd&td'd(gd)td*d+gd,td-d.gd/td0d1gd2td3d4gd5td6d7gd8td9d:gd;td<d=gd>td?d@gdAtdBdCgdDtdEdFgdGtdHdIgdJtdKdLgdMtdNdOgdPtdQdRgdStdTdUgdVtdWdXgdYtdZd[gd\td]d^gd_td`dagdbtdcddgdetdfdggdhtdidjgdktdldmgdntdodpgdqtdrdsgdttdudvgdwtdxdygdztd{d|gd}td~dgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdtddgdgH|_tj|||dS)zInitialize the class.z -sequencesequencez Sequence to choose primers from. is_requiredTz-taskZtaskz#Tell eprimer3 what task to perform.z -hybridprobeZ hybridprobez-Find an internal oligo to use as a hyb probe.z -numreturnZ numreturnz)Maximum number of primer pairs to return.z-includedregionZincludedregionz3Subregion of the sequence in which to pick primers.z-targettargetz(Sequence to target for flanking primers.z-excludedregionZexcludedregionz'Regions to exclude from primer picking.z -forwardinputZ forwardinputz&Sequence of a forward primer to check.z -reverseinputZ reverseinputz&Sequence of a reverse primer to check.z-gcclampZgcclampz:The required number of Gs and Cs at the 3' of each primer.z-osizeZosizez!Optimum length of a primer oligo.z-minsizeZminsizez!Minimum length of a primer oligo.z-maxsizemaxsizez!Maximum length of a primer oligo.z-otmZotmzMelting temperature for primer oligo (OBSOLETE). Option replaced in EMBOSS 6.6.0 by -opttm z-opttmZopttmzOptimum melting temperature for a primer oligo. Option added in EMBOSS 6.6.0, replacing -otm z-mintmZmintmz/Minimum melting temperature for a primer oligo.z-maxtmZmaxtmz/Maximum melting temperature for a primer oligo.z -maxdifftmZ maxdifftmzOMaximum difference in melting temperatures between forward and reverse primers.z -ogcpercentZ ogcpercentzOptimum GC% for a primer.z-mingcZmingczMinimum GC% for a primer.z-maxgcZmaxgczMaximum GC% for a primer.z -saltconcZsaltconcz)Millimolar salt concentration in the PCR.z-dnaconcZdnaconcz7Nanomolar concentration of annealing oligos in the PCR.z -maxpolyxZmaxpolyxz;Maximum allowable mononucleotide repeat length in a primer.z -productosizeZ productosizezOptimum size for the PCR product (OBSOLETE). Option replaced in EMBOSS 6.1.0 by -psizeopt z -psizeoptZpsizeoptzOptimum size for the PCR product. Option added in EMBOSS 6.1.0, replacing -productosize z-productsizerangeZproductsizerangezAcceptable range of length for the PCR product (OBSOLETE). Option replaced in EMBOSS 6.1.0 by -prange z-prangeZprangezAcceptable range of length for the PCR product. Option added in EMBOSS 6.1.0, replacing -productsizerange z -productotmZ productotmzOptimum melting temperature for the PCR product (OBSOLETE). Option replaced in EMBOSS 6.1.0 by -ptmopt z-ptmoptZptmoptzOptimum melting temperature for the PCR product. Option added in EMBOSS 6.1.0, replacing -productotm z -productmintmZ productmintmzMinimum allowed melting temperature for the amplicon (OBSOLETE) Option replaced in EMBOSS 6.1.0 by -ptmmin z-ptmminZptmminzMinimum allowed melting temperature for the amplicon."), Option added in EMBOSS 6.1.0, replacing -productmintm z -productmaxtmZ productmaxtmzMaximum allowed melting temperature for the amplicon (OBSOLETE). Option replaced in EMBOSS 6.1.0 by -ptmmax z-ptmmaxZptmmaxzMaximum allowed melting temperature for the amplicon."), Option added in EMBOSS 6.1.0, replacing -productmaxtm z-oexcludedregionZoexcludedregionzDo not pick internal oligos in this region."), Option added in EMBOSS 6.1.0, replacing -oligoexcludedregion. z-oligoexcludedregionZoligoexcludedregionzDo not pick internal oligos in this region (OBSOLETE)."), Option replaced in EMBOSS 6.1.0 by -oexcluderegion. z -oligoinputZ oligoinputzSequence of the internal oligo.z -oligosizeZ oligosizezOptimum length of internal oligo (OBSOLETE). Option replaced in EMBOSS 6.1.0 by -osizeopt. z -osizeoptZosizeoptz}Optimum length of internal oligo. Option added in EMBOSS 6.1.0, replaces -oligosize z -oligominsizeZ oligominsizezMinimum length of internal oligo (OBSOLETE)."), Option replaced in EMBOSS 6.1.0 by -ominsize. z -ominsizeZominsizezMinimum length of internal oligo." Option added in EMBOSS 6.1.0, replaces -oligominsize z -oligomaxsizeZ oligomaxsizezMaximum length of internal oligo (OBSOLETE). Option replaced in EMBOSS 6.1.0 by -omaxsize. z -omaxsizeZomaxsizezMaximum length of internal oligo. Option added in EMBOSS 6.1.0, replaces -oligomaxsize z-oligotmZoligotmzOptimum melting temperature of internal oligo (OBSOLETE). Option replaced in EMBOSS 6.1.0 by -otmopt. z-otmoptZotmoptzvOptimum melting temperature of internal oligo. Option added in EMBOSS 6.1.0. z -oligomintmZ oligomintmzMinimum melting temperature of internal oligo (OBSOLETE). Option replaced in EMBOSS 6.1.0 by -otmmin. z-otmminZotmminzMinimum melting temperature of internal oligo. Option added in EMBOSS 6.1.0, replacing -oligomintm z -oligomaxtmZ oligomaxtmzMaximum melting temperature of internal oligo (OBSOLETE). Option replaced in EMBOSS 6.1.0 by -otmmax. z-otmmaxZotmmaxzMaximum melting temperature of internal oligo. Option added in EMBOSS 6.1.0, replacing -oligomaxtm z-oligoogcpercentZoligoogcpercentzOptimum GC% for internal oligo (OBSOLETE). Option replaced in EMBOSS 6.1.0 by -ogcopt. z-ogcoptZogcoptzOptimum GC% for internal oligo." Option added in EMBOSS 6.1.0, replacing -oligoogcpercent z -oligomingcZ oligomingczMinimum GC% for internal oligo (OBSOLETE). Option replaced in EMBOSS 6.1.0 by -ogcmin. z-ogcminZogcminz}Minimum GC% for internal oligo. Option added in EMBOSS 6.1.0, replacing -oligomingc z -oligomaxgcZ oligomaxgcztMaximum GC% for internal oligo. Option replaced in EMBOSS 6.1.0 by -ogcmax z-ogcmaxZogcmaxzMaximum GC% for internal oligo."), Option added in EMBOSS 6.1.0, replacing -oligomaxgc z-oligosaltconcZ oligosaltconczMillimolar concentration of salt in the hybridisation."), Option replaced in EMBOSS 6.1.0 by -osaltconc z -osaltconcZ osaltconczMillimolar concentration of salt in the hybridisation."), Option added in EMBOSS 6.1.0, replacing -oligosaltconc z -oligodnaconcZ oligodnaconczNanomolar concentration of internal oligo in the hybridisation. Option replaced in EMBOSS 6.1.0 by -odnaconc z -odnaconcZodnaconczNanomolar concentration of internal oligo in the hybridisation. Option added in EMBOSS 6.1.0, replacing -oligodnaconc z -oligoselfanyZ oligoselfanyzMaximum allowable alignment score for self-complementarity (OBSOLETE). Option replaced in EMBOSS 6.1.0 by -oanyself z -oanyselfZoanyselfzMaximum allowable alignment score for self-complementarity."), Option added in EMBOSS 6.1.0, replacing -oligoselfany z -oligoselfendZ oligoselfendzMaximum allowable 3'-anchored global alignment score " for self-complementarity (OBSOLETE). Option replaced in EMBOSS 6.1.0 by -oendself z -oendselfZoendselfzMax 3'-anchored self-complementarity global alignment score. Option added in EMBOSS 6.1.0, replacing -oligoselfend z-oligomaxpolyxZ oligomaxpolyxzMaximum length of mononucleotide repeat in internal oligo (OBSOLETE). Option replaced in EMBOSS 6.1.0 by -opolyxmax z -opolyxmaxZ opolyxmaxzMaximum length of mononucleotide repeat in internal oligo."), Option added in EMBOSS 6.1.0, replacing -oligomaxpolyx z-mispriminglibraryfileZmispriminglibraryfilez8File containing library of sequences to avoid amplifyingz-maxmisprimingZ maxmisprimingz]Maximum allowed similarity of primers to sequences in library specified by -mispriminglibraryz-oligomaxmishybZoligomaxmishybzMaximum alignment score for hybridisation of internal oligo to library specified by -oligomishyblibraryfile (OBSOLETE). Option replaced in EMBOSS 6.1.0 by -omishybmax z -omishybmaxZ omishybmaxzMaximum alignment score for hybridisation of internal oligo to library specified by -mishyblibraryfile. Option added in EMBOSS 6.1.0, replacing -oligomaxmishyb z-oligomishyblibraryfileZoligomishyblibraryfilezLibrary file of seqs to avoid internal oligo hybridisation (OBSOLETE). Option replaced in EMBOSS 6.1.0 by -mishyblibraryfile z-mishyblibraryfileZmishyblibraryfilezLibrary file of seqs to avoid internal oligo hybridisation. Option added in EMBOSS 6.1.0, replacing -oligomishyblibraryfile z -explainflagZ explainflagz,Produce output tags with eprimer3 statisticsN)rrrr)rrrrrrrs$                                                                        zPrimer3Commandline.__init__N)rrrrrrrrrr"s r"c@s%eZdZdZdddZdS)PrimerSearchCommandlinez 0.001)z -invarfracZ invarfraczproportoin of invariant sitesz-ttratiottratioz ts/tv ratioz -freqsfromZ freqsfromzuse emprical base freqsz -basefreqbasefreqzspecify basefreqsz-lowerlowerzlower triangle matrix (y/N)N)rrrr)rrrrrrrs<             zFDNADistCommandline.__init__N)rrrrrrrrrr,s r,c@s%eZdZdZdddZdS)FTreeDistCommandlinezCommandline object for the ftreedist program from EMBOSS. ftreedist is an EMBOSS wrapper for the PHYLIP program treedist used for calulating distance measures between phylogentic trees. Z ftreedistcKstddgdddddtddgd td d gd td dgdtddgdtddgdg|_tj|||dS)zInitialize the class.z -intreefile intreefileztree file to score (phylip)rTr$z-dtypeZdtypez*distance type ([S]ymetric, [b]ranch score)z-pairingZpairingzzWrite trees to file (Y/n)z -outtreefiler?zfilename for output treez-dotdiffrEzUse dot-differencing? [Y/n]N)rrrr)rrrrrrrsB               zFDNAParsCommandline.__init__N)rrrrrrrrrrFs rFc@s%eZdZdZdddZdS)FProtParsCommandlineaCommandline object for the fdnapars program from EMBOSS. fprotpars is an EMBOSS version of the PHYLIP program protpars, for estimating trees from protein sequences using parsiomny. Calling this command without providing a value for the option "-intreefile" will invoke "interactive mode" (and as a result fail if called with subprocess) if "-auto" is not set to true. Z fprotparscKstddgdddddtddgd td d gd ddddtd dgdtddgdtddgdtddgdtddgdtddgdtdd gd!td"d#gd$td%d&gd'g |_tj|||d(S))zInitialize the class.z -sequencer#zseq file to use (phylip)rTr$z -intreefiler8zPhylip tree file to scorez -outtreefiler?zphylip tree output filez-weightsr2z weights filez -whichcode whichcodez which genetic code, [U,M,V,F,Y]]z-njumblerGz7number of times to randomise input order (default is 0)z-seedr=zprovide random seedz-outgrnor:zSpecify outgroupz-threshrHzUse threshold parsimony (y/N)z -thresholdrIzThreshold valuez-troutr>zWrite trees to file (Y/n)z-dotdiffrEzUse dot-differencing? [Y/n]N)rrrr)rrrrrrrs:            zFProtParsCommandline.__init__N)rrrrrrrrrrJs rJc@s%eZdZdZdddZdS)FProtDistCommandlinezCommandline object for the fprotdist program from EMBOSS. fprotdist is an EMBOSS wrapper for the PHYLIP program protdist used to estimate trees from protein sequences using parsimony Z fprotdistcKs(tddgdddddtddgd td d gd td dgdtddgdtddgdtddgdtddgdtddgdtdd gd!td"d#gd$td%d&gd'td(d)gd*td+d,gd-g|_tj|||d.S)/zInitialize the class.z -sequencer#zseq file to use (phylip)rTr$z -ncategoriesr/z number of rate catergories (1-9)z-rater0zrate for each categoryz -catergoriesrBz file of ratesz-weightsr2z weights filez-methodr-zsub. model [j,h,d,k,s,c]z-gammar.zgamma [g, i,c]z-gammacoefficientr3zvalue for gamma (> 0.001)z-invarcoefficientZinvarcoefficientz4float for variation of substitution rate among sitesz-aacategZaacategz"Choose the category to use [G,C,H]z -whichcoderKzgenetic code [c,m,v,f,y]z-easeZeasez-Pob change catergory (float between -0 and 1)z-ttratior4z#Transition/transversion ratio (0-1)z -basefreqr5z+DNA base frequencies (space separated list)N)rrrr)rrrrrrrs>              zFProtDistCommandline.__init__N)rrrrrrrrrrLs rLc@s%eZdZdZdddZdS)FConsenseCommandlinezCommandline object for the fconsense program from EMBOSS. fconsense is an EMBOSS wrapper for the PHYLIP program consense used to calculate consensus trees. Z fconsensec Kstddgdddddtddgd td d gd td dgdtddgdtddgdtddgdg|_tj|||dS)zInitialize the class.z -intreefiler8z-file with phylip trees to make consensus fromrTr$z-methodr-z!consensus method [s, mr, MRE, ml]z-mlfracZmlfracz8cut-off freq for branch to appear in consensus (0.5-1.0)z-rootrootztreat trees as rooted (YES, no)z-outgrnor:z&OTU to use as outgroup (starts from 0)z-troutr>z -outtreefiler?z"Phylip tree output file (optional)N)rrrr)rrrrrrrs"       zFConsenseCommandline.__init__N)rrrrrrrrrrM s rMc@s%eZdZdZdddZdS)WaterCommandlinez5Commandline object for the water program from EMBOSS.Zwaterc Kstddgdddddtddgd ddddtd d gd ddtd dgdddtddgdddtddgdtddgdtddgdtddgdtdd gd!td"d#gd$g |_tj|||d%S)&zInitialize the class.z -asequence asequencezFirst sequence to alignrTr$z -bsequence bsequencezSecond sequence to alignz-gapopengapopenzGap open penaltyz -gapextend gapextendzGap extension penaltyz -datafiler<z Matrix filez-nobriefnobriefz(Display extended identity and similarityz-briefbriefz%Display brief identity and similarityz -similarity similarityz'Display percent identity and similarityz -snucleotider*z"Sequences are nucleotide (boolean)z -sproteinr+zSequences are protein (boolean)z-aformataformatz5Display output in a different specified output formatN)rrrrr)rrrrrrr-s<           zWaterCommandline.__init__N)rrrrrrrrrrO*s rOc@s%eZdZdZdddZdS)NeedleCommandlinez6Commandline object for the needle program from EMBOSS.ZneedlecKsFtddgdddddtddgd ddddtd d gd ddtd dgdddtddgdddtddgdtddgdtddgdtddgdtdd gd!td"d#gd$td%d&gd'td(d)gd*td+d,gd-g|_tj|||d.S)/zInitialize the class.z -asequencerPzFirst sequence to alignrTr$z -bsequencerQzSecond sequence to alignz-gapopenrRzGap open penaltyz -gapextendrSzGap extension penaltyz -datafiler<z Matrix filez -endweight endweightzApply And gap penaltiesz-endopenendopenz0The score taken away when an end gap is created.z -endextend endextendzPThe score added to the end gap penality for each base or residue in the end gap.z-nobriefrTz(Display extended identity and similarityz-briefrUz%Display brief identity and similarityz -similarityrVz'Display percent identity and similarityz -snucleotider*z"Sequences are nucleotide (boolean)z -sproteinr+zSequences are protein (boolean)z-aformatrWz5Display output in a different specified output formatN)rrrrr)rrrrrrrTsH              zNeedleCommandline.__init__N)rrrrrrrrrrXQs rXc@s%eZdZdZdddZdS)NeedleallCommandlinez9Commandline object for the needleall program from EMBOSS.Z needleallcKsjtddgdddddtddgd ddddtd d gd ddtd dgdddtddgdddtddgdtddgdtddgdtddgdtdd gd!td"d#gd$td%d&gd'td(d)gd*td+d,gd-td.d/gd0td1d2gd3g|_tj|||d4S)5zInitialize the class.z -asequencerPzFirst sequence to alignrTr$z -bsequencerQzSecond sequence to alignz-gapopenrRzGap open penaltyz -gapextendrSzGap extension penaltyz -datafiler<z Matrix filez -minscoreminscorez:Exclude alignments with scores below this threshold score.z -errorfileZ errorfilezError file to be written to.z -endweightrYzApply And gap penaltiesz-endopenrZz0The score taken away when an end gap is created.z -endextendr[zPThe score added to the end gap penality for each base or residue in the end gap.z-nobriefrTz(Display extended identity and similarityz-briefrUz%Display brief identity and similarityz -similarityrVz'Display percent identity and similarityz -snucleotider*z"Sequences are nucleotide (boolean)z -sproteinr+zSequences are protein (boolean)z-aformatrWz5Display output in a different specified output formatN)rrrrr)rrrrrrrsP                zNeedleallCommandline.__init__N)rrrrrrrrrr\s r\c@s%eZdZdZdddZdS)StretcherCommandlinez9Commandline object for the stretcher program from EMBOSS.Z stretcherc Kstddgdddddtddgd ddddtd d gd ddd ddtddgdddd ddtddgdddtddgdtddgdtddgdg|_tj|||d S)!zInitialize the class.z -asequencerPzFirst sequence to alignrTr$z -bsequencerQzSecond sequence to alignz-gapopenrRzGap open penaltyZchecker_functioncSs t|tS)N) isinstanceint)valuerrrsz/StretcherCommandline.__init__..z -gapextendrSzGap extension penaltycSs t|tS)N)r_r`)rarrrrbsz -datafiler<z Matrix filez -snucleotider*z"Sequences are nucleotide (boolean)z -sproteinr+zSequences are protein (boolean)z-aformatrWz5Display output in a different specified output formatN)rrrr)rrrrrrrs4          zStretcherCommandline.__init__N)rrrrrrrrrr^s r^c@s%eZdZdZdddZdS)FuzznucCommandlinez7Commandline object for the fuzznuc program from EMBOSS.ZfuzznuccKstddgdddtddgdddtd d gd td d gdtddgdg|_tj|||dS)zInitialize the class.z -sequencer#zSequence database USAr$Tz-patternpatternz5Search pattern, using standard IUPAC one-letter codesz -pmismatch pmismatchzNumber of mismatchesz -complementZ complementzSearch complementary strandz-rformatrformatz'Specify the report format to output in.N)rrrr)rrrrrrrs     zFuzznucCommandline.__init__N)rrrrrrrrrrcs rcc@s%eZdZdZdddZdS)FuzzproCommandlinez7Commandline object for the fuzzpro program from EMBOSS.ZfuzzprocKsttddgdddtddgdddtd d gd td d gdg|_tj|||dS)zInitialize the class.z -sequencer#zSequence database USAr$Tz-patternrdz5Search pattern, using standard IUPAC one-letter codesz -pmismatchrezNumber of mismatchesz-rformatrfz'Specify the report format to output in.N)rrrr)rrrrrrrs    zFuzzproCommandline.__init__N)rrrrrrrrrrgs rgc@s%eZdZdZdddZdS)Est2GenomeCommandlinez:Commandline object for the est2genome program from EMBOSS.Z est2genomecKs^tddgdddtddgdddtd d gd td d gdtddgdtddgdtddgdtddgdtddgdtddgd td!d"gd#td$d%gd&td'd(gd)td*d+gd,td-d.gd/td0d1gd2td3d4gd5g|_tj|||d6S)7zInitialize the class.z-estZestzEST sequence(s)r$Tz-genomeZgenomezGenomic sequencez-matchmatchzScore for matching two basesz -mismatchmismatchzCost for mismatching two basesz -gappenaltyZ gappenaltyzECost for deleting a single base in either sequence, excluding intronsz-intronpenaltyZ intronpenaltyz*Cost for an intron, independent of length.z-splicepenaltyZ splicepenaltyzUCost for an intron, independent of length and starting/ending on donor-acceptor sitesz -minscorer]z:Exclude alignments with scores below this threshold score.z-reversereversez+Reverse the orientation of the EST sequencez-spliceZsplicez$Use donor and acceptor splice sites.z-modemodez?This determines the comparion mode. 'both', 'forward' 'reverse'z-bestbestz:You can print out all comparisons instead of just the bestz-spacespacezfor linear-space recursion.z-shuffleshuffleZShufflez-seedr=zRandom number seedz-alignZalignzShow the alignment.z-widthwidthzAlignment widthN)rrrr)rrrrrrrsJ                 zEst2GenomeCommandline.__init__N)rrrrrrrrrrhs rhc@s%eZdZdZdddZdS)ETandemCommandlinez7Commandline object for the etandem program from EMBOSS.Zetandemc Kstddgdddddtddgd ddtd d gd ddtd dgdtddgdtddgdtddgdg|_tj|||dS)zInitialize the class.z -sequencer#SequencerTr$z -minrepeatZ minrepeatzMinimum repeat sizez -maxrepeat maxrepeatzMaximum repeat sizez -thresholdrIzThreshold scorez -mismatchrjzAllow N as a mismatchz-uniformuniformzAllow uniform consensusz-rformatrfzOutput report formatN)rrrr)rrrrrrr2s&       zETandemCommandline.__init__N)rrrrrrrrrrq/s rqc@s%eZdZdZdddZdS)EInvertedCommandlinez9Commandline object for the einverted program from EMBOSS.Z einvertedc Kstddgdddddtddgd ddddtd d gd ddtd dgdddtddgdddtddgdg|_tj|||dS)zInitialize the class.z -sequencer#rrrTr$z-gapZgapz Gap penaltyz -thresholdrIzMinimum score thresholdz-matchriz Match scorez -mismatchrjzMismatch scorez -maxrepeatrsz6Maximum separation between the start and end of repeatN)rrrr)rrrrrrrNs(      zEInvertedCommandline.__init__N)rrrrrrrrrruKs ruc@s%eZdZdZdddZdS)PalindromeCommandlinez:Commandline object for the palindrome program from EMBOSS.Z palindromec Kstddgdddddtddgd ddtd d gd ddtd dgdddtddgdddtddgdddg|_tj|||dS)zInitialize the class.z -sequencer#rrrTr$z -minpallenZ minpallenzMinimum palindrome lengthz -maxpallenZ maxpallenzMaximum palindrome lengthz -gaplimitZgaplimitzMaximum gap between repeatsz-nummismatchesZ nummismatcheszNumber of mismatches allowedz-overlapoverlapzReport overlapping matchesN)rrrr)rrrrrrrks(      zPalindromeCommandline.__init__N)rrrrrrrrrrvhs rvc@s%eZdZdZdddZdS)TranalignCommandlinez9Commandline object for the tranalign program from EMBOSS.Z tranalignc Kstddgdddddtddgd ddddtd d gd ddddtd dgdg|_tj|||dS)zInitialize the class.z -asequencerPz#Nucleotide sequences to be aligned.rTr$z -bsequencerQzProtein sequence alignmentz-outseqoutseqzOutput sequence file.z-tabletablez Code to useN)rrrr)rrrrrrrs    zTranalignCommandline.__init__N)rrrrrrrrrrxs rxc@s%eZdZdZdddZdS)DiffseqCommandlinez7Commandline object for the diffseq program from EMBOSS.Zdiffseqc Kstddgdddddtddgd ddddtd d gd ddtd dgdddddtddgdddddtddgdg|_tj|||dS)zInitialize the class.z -asequencerPzFirst sequence to comparerTr$z -bsequencerQzSecond sequence to comparez -wordsizewordsizez-Word size to use for comparisons (10 default)z -aoutfeatZaoutfeatz,File for output of first sequence's featuresz -boutfeatZboutfeatz-File for output of second sequence's featuresz-rformatrfzOutput report file formatN)rrrr)rrrrrrrs,      zDiffseqCommandline.__init__N)rrrrrrrrrr{s r{c@s%eZdZdZdddZdS)IepCommandlineaCommandline for EMBOSS iep: calculated isoelectric point and charge. Examples -------- >>> from Bio.Emboss.Applications import IepCommandline >>> iep_cline = IepCommandline(sequence="proteins.faa", ... outfile="proteins.txt") >>> print(iep_cline) iep -outfile=proteins.txt -sequence=proteins.faa You would typically run the command line with iep_cline() or via the Python subprocess module, as described in the Biopython tutorial. ZiepcKstddgdddddtddgd td d gd td dgdtddgdtddgdg|_tj|||dS)zInitialize the class.z -sequencer#zProtein sequence(s) filenamerTr$z-aminoZaminozZNumber of N-termini Integer 0 (default) or more. z -carboxylZcarboxylzZNumber of C-termini Integer 0 (default) or more. z-lysinemodifiedZlysinemodifiedzaNumber of modified lysines Integer 0 (default) or more. z -disulphidesZ disulphideszcNumber of disulphide bridges Integer 0 (default) or more. z -noterminiZ noterminiz=Exclude (True) or include (False) charge at N and C terminus.N)rrrr)rrrrrrrs      zIepCommandline.__init__N)rrrrrrrrrr}s r}c@s1eZdZdZdddZddZdS)SeqretCommandlineaCommandline object for the seqret program from EMBOSS. This tool allows you to interconvert between different sequence file formats (e.g. GenBank to FASTA). Combining Biopython's Bio.SeqIO module with seqret using a suitable intermediate file format can allow you to read/write to an even wider range of file formats. This wrapper currently only supports the core functionality, things like feature tables (in EMBOSS 6.1.0 onwards) are not yet included. ZseqretcKsttddgdddtddgdddtd d gd td d gdg|_tj|||dS)zInitialize the class.z -sequencer#zInput sequence(s) filenamerTz-outseqryzOutput sequence file.z-sformatZsformatz.Input sequence(s) format (e.g. fasta, genbank)z -osformatZosformatz/Output sequence(s) format (e.g. fasta, genbank)N)rrrr)rrrrrrrs    zSeqretCommandline.__init__cCs[|jp|jp|js'td|jp?|jp?|jsNtdtj|S)Nz]You must either set outfile (output filename), or enable filter or stdout (output to stdout).z\You must either set sequence (input filename), or enable filter or stdin (input from stdin).)ryrrr r#Zstdintrr!)rrrrr! s   zSeqretCommandline._validateN)rrrrrr!rrrrr~s r~c@s%eZdZdZdddZdS)SeqmatchallCommandlineakCommandline object for the seqmatchall program from EMBOSS. e.g. >>> cline = SeqmatchallCommandline(sequence="opuntia.fasta", outfile="opuntia.txt") >>> cline.auto = True >>> cline.wordsize = 18 >>> cline.aformat = "pair" >>> print(cline) seqmatchall -auto -outfile=opuntia.txt -sequence=opuntia.fasta -wordsize=18 -aformat=pair Z seqmatchallcKsbtddgdddddtddgd td d gd g|_tj|||d S)zInitialize the class.z -sequencer#zReadable set of sequencesrTr$z -wordsizer|z(Word size (Integer 2 or more, default 4)z-aformatrWz5Display output in a different specified output formatN)rrrr)rrrrrrr%s   zSeqmatchallCommandline.__init__N)rrrrrrrrrrs r__main__) run_doctestN)%r __future__rZBio.Applicationrrrrrr"r'r,r7r;rArFrJrLrMrOrXr\r^rcrgrhrqrurvrxr{r}r~rrZ Bio._utilsrrrrrsBF$Q+$*1-,'.2#13)