:>"^@sqdZddlmZddlmZmZddlZGdddZGdddeZ d d Z dS) zParse XMS motif files.)motifs)minidomNodeNc@sReZdZdZddZddZddZdd Zd d Zd S) XMSScannerz Class for scanning XMS XML file.cCsIt|_x6|jdD]%}|jtjkr|j|qWdS)zDGenerate motif Record from xms document, an XML-like motif pfm file.motifN)RecordrecordgetElementsByTagNamenodeTyper ELEMENT_NODE handle_motif)selfdocchildr3/tmp/pip-build-ww9dw3qa/biopython/Bio/motifs/xms.py__init__s zXMSScanner.__init__cs|j|jd}dgdgdgdgixL|jdD];}fddtddddg|j|DqFWtjd d d }||_|jj|d S) zHRead the motif's name and column from the node and add the motif record.nameACGTcolumncs/g|]%\}}|jt|qSr)appendfloat).0Z nucleotideZnucleotide_count)nucleotide_countsrr s z+XMSScanner.handle_motif..ZalphabetZGATCcountsN) get_textr zipget_acgtrZMotifrrr)r nodeZ motif_namerrr)rrr s* zXMSScanner.handle_motifcCsx|jdD]}d}d}xt|jD]i}|jtjkrGq,|jdkrt|j|g|krtd}|jdkr,|j|g}q,W|r|SqWdS)zCExtract the value of the motif's property named key_name from node.propFNkeyTvalue)r childNodesr rr ZtagNamer)r r"Zkey_nameZ cur_propertyZright_propertyZ cur_valuerrrrget_property_value(s'zXMSScanner.get_property_valuecCsd\}}}}x|jdD]}|jddkrXt|j|g}q"|jddkrt|j|g}q"|jddkrt|j|g}q"|jddkr"t|j|g}q"W||||fS) z1Get and return the motif's weights of A, C, G, T.weightsymbolZadenineZcytosineZguanineZthymine)r(r(r(r()r getAttributerr)r r"acgtr)rrrr!8szXMSScanner.get_acgtcCszg}xU|D]M}|jtjkr8|j|jq |jr |j|j|jq Wtj dddj |S)zMReturn a string representation of the motif's properties listed on nodelist .z\s+ ) r rZ TEXT_NODErZ wholeTextZ hasChildNodesrr&resubjoin)r ZnodelistZretlistr"rrrrFs  zXMSScanner.get_textN) __name__ __module__ __qualname____doc__rr r'r!rrrrrrs     rc@s"eZdZdZddZdS)rzClass to store the information in a XMS matrix table. The record inherits from a list containing the individual motifs. cCsdjdd|DS)N css|]}t|VqdS)N)str)rrrrr Ysz!Record.__str__..)r4)r rrr__str__XszRecord.__str__N)r5r6r7r8r<rrrrrRs rcCs"tj|}t|j}|S)zRead motifs in XMS matrix format from a file handle. XMS is an XML format for describing regulatory motifs and PSSMs. This format was defined by Thomas Down, and used in the NestedMICA and MotifExplorer programs. )rparserr)handleZxms_docrrrrread\sr?) r8ZBiorxml.domrrr2rlistrr?rrrrs  B