:>"^I@sdZddlmZmZddlmZddlmZddlm Z m Z m Z ddl m Z mZmZddlmZGdd d eZe d d d Ze d dZddZddZddZddZddZedkrddlmZedS)zCode for dealing with Codon Alignment. CodonAlignment class is inherited from MultipleSeqAlignment class. This is the core class to deal with codon alignment in biopython. )divisionprint_function)MultipleSeqAlignment) SeqRecord)default_codon_tabledefault_codon_alphabetcompare_codon_alphabet)_get_codon_listCodonSeq cal_dn_ds) chisqprobc@seZdZdZddeddZddZddd Zd d Zd d Z ddZ de ddZ dde ddZ eeddZdS)CodonAlignmentaCodon Alignment class that inherits from MultipleSeqAlignment. >>> from Bio.Alphabet import generic_dna >>> from Bio.SeqRecord import SeqRecord >>> from Bio.Alphabet import IUPAC, Gapped >>> a = SeqRecord(CodonSeq("AAAACGTCG", alphabet=default_codon_alphabet), id="Alpha") >>> b = SeqRecord(CodonSeq("AAA---TCG", alphabet=default_codon_alphabet), id="Beta") >>> c = SeqRecord(CodonSeq("AAAAGGTGG", alphabet=default_codon_alphabet), id="Gamma") >>> print(CodonAlignment([a, b, c])) CodonAlphabet(Standard) CodonAlignment with 3 rows and 9 columns (3 codons) AAAACGTCG Alpha AAA---TCG Beta AAAAGGTGG Gamma NcCsktj||d|x,|D]$}t|jtstdqW|jddkrgtddS)zInitialize the class.alphabetzACodonSeq objects are expected in each SeqRecord in CodonAlignmentrzKAlignment length is not a multiple of three (i.e. a whole number of codons)N)r__init__ isinstanceseqr TypeErrorget_alignment_length ValueError)selfrecordsnamerrecrB/tmp/pip-build-ww9dw3qa/biopython/Bio/codonalign/codonalignment.pyr(s  zCodonAlignment.__init__cstj}dtj|jjfg}|dkro|jfddjDn]|jfddjddD|jd|jjjd d dd j |S) aDReturn a multi-line string summary of the alignment. This output is indicated to be readable, but large alignment is shown truncated. A maximum of 20 rows (sequences) and 60 columns (20 codons) are shown, with the record identifiers. This should fit nicely on a single screen. e.g. z9%s CodonAlignment with %i rows and %i columns (%i codons)<cs%g|]}j|ddqS)lengthr) _str_line).0r)rrr Es z*CodonAlignment.__str__..cs%g|]}j|ddqS)rr)r)r r)rrrr!Hs Nz...r ) len_recordsstr _alphabetrget_aln_lengthextendappendrjoin)rZrowslinesr)rr__str__6s   #zCodonAlignment.__str__cs,t|tr|j|St|tr@t|j||jSt|dkr^td|\}t|tr|j|Sttrdjfdd|j|DSddl m }|dkrt fd d|j|D|St fd d|j|D|SdS) z3Return a CodonAlignment object for single indexing.zInvalid index type.rc3s|]}t|VqdS)N)r()r r) col_indexrr [sz-CodonAlignment.__getitem__..r)generic_nucleotideNc3s|]}|VqdS)Nr)r r)r1rrr2`sc3s|]}|VqdS)Nr)r r)r1rrr2ds) rintr'slicer r)r&rr-Z Bio.Alphabetr3r)rindexrZ row_indexr3r)r1r __getitem__Ns(    zCodonAlignment.__getitem__cCst|trt|t|kr3tdt|j|jr}|j}ddt||D}t|d|Stdn\t|trt|t|krtd|j |Stdj t |dS)ahCombine two codonalignments with the same number of rows by adding them. The method also allows to combine a CodonAlignment object with a MultipleSeqAlignment object. The following rules apply: * CodonAlignment + CodonAlignment -> CodonAlignment * CodonAlignment + MultipleSeqAlignment -> MultipleSeqAlignment zTWhen adding two alignments they must have the same length (i.e. same number or rows)cssL|]B\}}tdtt|jt|jd|jjVqdS)rrN)rr r(rr)r leftrightrrrr2xsz)CodonAlignment.__add__..rz@Only CodonAlignment with the same CodonAlphabet can be combined.zjOnly CodonAlignment or MultipleSeqAlignment object can be added with a CodonAlignment object. {} detected.N) rr r&rrr)ziprrtoMultipleSeqAlignmentformatobject)rotheralphaZmergedrrr__add__hs     zCodonAlignment.__add__cCs|jdS)zGet aligment length.r)r)rrrrr*szCodonAlignment.get_aln_lengthcCs dd|jD}t|S)zConvert the CodonAlignment to a MultipleSeqAlignment. Return a MultipleSeqAlignment containing all the SeqRecord in the CodonAlignment using Seq to store sequences cSs.g|]$}t|jjd|jqS)id)rrZtoSeqrA)r rrrrr!s z9CodonAlignment.toMultipleSeqAlignment..)r'r)rZ alignmentsrrrr;s  z%CodonAlignment.toMultipleSeqAlignmentZNG86c Cs=ddlm}dd|jD}t|j}g}g}xt|D]}|jg|jgxt|dD]} || krt|j||j| d|d|\} } ||j| ||j| q||jd||jdqWqNW||d |} ||d |} | | fS) zAvailable methods include NG86, LWL85, YN00 and ML. Argument: - method - Available methods include NG86, LWL85, YN00 and ML. - codon_table - Codon table to use for forward translation. r)DistanceMatrixcSsg|]}|jqSr)rA)r irrrr!s z3CodonAlignment.get_dn_ds_matrix..r#method codon_tablegZmatrix)Bio.Phylo.TreeConstructionrBr'r&ranger,r )rrDrEZDMnamessizeZ dn_matrixZ ds_matrixrCjdnZdsdn_dmds_dmrrrget_dn_ds_matrixs&   zCodonAlignment.get_dn_ds_matrixUPGMAc Csddlm}|jd|d|\}}|}|}|dkrm|j|} |j|} nB|dkr|j|} |j|} ntdj|| | fS)zCnstruct dn tree and ds tree. Argument: - dn_ds_method - Available methods include NG86, LWL85, YN00 and ML. - tree_method - Available methods include UPGMA and NJ. r)DistanceTreeConstructorrDrErOZNJz:Unknown tree method ({0}). Only NJ and UPGMA are accepted.)rFrPrNZupgmaZnj RuntimeErrorr<) rZ dn_ds_methodZ tree_methodrErPrLrMZdn_constructorZds_constructorZdn_treeZds_treerrrget_dn_ds_trees      zCodonAlignment.get_dn_ds_treecs,fdd|jD}||dS)zConvert a MultipleSeqAlignment to CodonAlignment. Function to convert a MultipleSeqAlignment to CodonAlignment. It is the user's responsibility to ensure all the requirement needed by CodonAlignment is met. cs:g|]0}ttt|jdd|jqS)rrA)rr r(rrA)r rC)rrrr!s z+CodonAlignment.from_msa..r)r')clsZalignrrr)rrfrom_msas zCodonAlignment.from_msa)__name__ __module__ __qualname____doc__rrr/r7r@r*r;rrNrR classmethodrTrrrrr s     r g?csddl}tdd|Ds1tddd|D}tt|dkrhtd |dd }|j|j}x|jD]}d ||} |j gx(| D] |dj t j qWqWg} x^t |D]Pg} x4|D],} fd d | D} | j | qW| j | qWd\}}}}t d|\}}x| D]djdd}d|kst|dkrqtddD}|rDt||}t||}t||}t|||}||7}||7}qt||}t||}t||}t|||}||7}||7}qWt||||gS)aMcDonald-Kreitman test for neutrality. Implement the McDonald-Kreitman test for neutrality (PMID: 1904993) This method counts changes rather than sites (http://mkt.uab.es/mkt/help_mkt.asp). Arguments: - codon_alns - list of CodonAlignment to compare (each CodonAlignment object corresponds to gene sampled from a species) Return the p-value of test result. rNcss|]}t|tVqdS)N)rr )r rCrrrr2szmktest..z#mktest accepts CodonAlignment list.cSsg|]}|jqSr)r)r rCrrrr!s zmktest..r#z;CodonAlignment object for mktest should be of equal length.rstopcsh|]}|qSrr)r k)rCrr s zmktest..rE-css!|]}t|dkVqdS)r#N)r&)r r[rrrr2sr%)rrrr)copyallrr&setrQdeepcopy forward_table stop_codonsr,r rrG_get_codon2codon_matrixunion _get_subgraph_count_replacement_G_test)Z codon_alnsrEr?r^Z codon_aln_lenZ codon_num codon_dictrZZ codon_lstZ codon_aln codon_setZ uniq_codonsrJZ uniq_codonZsyn_fixZ nonsyn_fixZsyn_polyZ nonsyn_polyGnonsyn_GZ all_codonZ fix_or_notZnonsyn_subgraphsubgraphZthis_nonZthis_synr)rCrmktestsV      "     rncCsd }ddt|jj|jD}|j}x|jD]}d||.rZrr#Ng?)rorprqrk)listrbkeysrcr& enumerate _dijkstra)rEZ base_tuplecodonsrirZnumrkrlgraphZ graph_nonsynrCZcodonpbrJZ tmp_codonZcodon1Zcodon2rrrrdsD     &       "rdcCsi}i}x(|jD]}d||(<) D[node] + child_value: is fixed. This function will return the distance between start and end. Arguments: - graph: Dictionary of dictionary (keys are vertices). - start: Start vertex. - end: End vertex. Output: List of vertices from the beginning to the end. drrNr#)rtrsr&removeitemscountinsertrG)rystartendDPnodeZ unseen_nodesZshortestZ temp_nodeZ child_nodeZ child_valuepathZdistancerCrrrrvEsH           "rvcCsxddlm}t|dkr&dSt|dkr^t|}|||d|dSt|}t|SdS)z9Count replacement needed for a given codon_set (PRIVATE).r)floorr#r0N)rr)mathrr&rs_prim)rjrkrrwrrrrgs  rgcCsddlm}ddlm}ddlm}m}m}g}g}x|jD]}|j |xj||D]^} || ||| f|krs| |||| f|krs|j || ||| fqsWqUW|t } xK|D]C\} } } | | j | | | f| | j | | | fqWg}t |d}| |ddd}||x|r||\}} } | |krn|j | |j | | |fx/| | D]#}|d|kr|||qWqnWd}x"|D]}|||d7}qW|S)zPrim's algorithm to find minimum spanning tree (PRIVATE). Code is adapted from http://programmingpraxis.com/2010/04/09/minimum-spanning-tree-prims-algorithm/ r)r) defaultdict)heapifyheappopheappushNr0) rr collectionsrheapqrrrrtr,rsr`add)rkrrrrrZnodesedgesrCrJconnZn1Zn2cZmstusedZ usable_edgesZcosterrzrrrrs< :&      rcCsXi}xK|D]C}i||>> print("%0.6f" % _G_test([17, 7, 42, 2])) 0.004924 r)logr0r#rN)rrsumr:r ) Z site_countsrrkZtotZtot_synZtot_nonZtot_fixZtot_polyexpZobsexrrrrhs   rh__main__) run_doctestN)rX __future__rrZ Bio.AlignrZ Bio.SeqRecordrZBio.codonalign.codonalphabetrrrZBio.codonalign.codonseqr r r ZBio.codonalign.chisqr r rnrdrvrgrrfrhrUZ Bio._utilsrrrrr s"@. D &