o Rŀgf@sZdZddlZddlmZddlmZddlmZGdddZGdd d eZ d d Z dS) zParse XMS motif files.N)minidom)Node)motifsc@s8eZdZdZddZddZddZdd Zd d Zd S) XMSScannerz Class for scanning XMS XML file.cCs2t|_|dD] }|jtjkr||q dS)zDGenerate motif Record from xms document, an XML-like motif pfm file.motifN)RecordrecordgetElementsByTagNamenodeTyper ELEMENT_NODE handle_motif)selfdocchildrB/var/www/html/myenv/lib/python3.10/site-packages/Bio/motifs/xms.py__init__s   zXMSScanner.__init__cst||d}ggggd|dD]}fddtgd||Dqtjdd}||_|j|dS) zHRead the motif's name and column from the node and add the motif record.name)ACGTcolumncs"g|] \}}|t|qSr)appendfloat).0 nucleotidenucleotide_countnucleotide_countsrr sz+XMSScanner.handle_motif..GATC)alphabetcountsN) get_textr zipget_acgtrMotifrrr)r node motif_namerrrrrr s zXMSScanner.handle_motifcCsv|dD]3}d}d}|jD]#}|jtjkrq|jdkr&||g|kr&d}|jdkr1||g}q|r8|SqdS)zCExtract the value of the motif's property named key_name from node.propFNkeyTvalue)r childNodesr rr tagNamer$)r r(key_name cur_propertyright_property cur_valuerrrrget_property_value,s    zXMSScanner.get_property_valuecCsd\}}}}|dD]A}|ddkrt||g}q |ddkr-t||g}q |ddkr=t||g}q |ddkrLt||g}q ||||fS)z1Get and return the motif's weights of A, C, G, T.)r4r4r4weightsymboladeninecytosineguaninethymine)r getAttributerr$)r r(acgtr5rrrr&<s  zXMSScanner.get_acgtcCsTg}|D]}|jtjkr||jq|jr|||jqt ddd |S)zMReturn a string representation of the motif's properties listed on nodelist .z\s+ ) r r TEXT_NODEr wholeText hasChildNodesr$r-resubjoin)r nodelistretlistr(rrrr$Js zXMSScanner.get_textN) __name__ __module__ __qualname____doc__rr r3r&r$rrrrrs rc@seZdZdZddZdS)rzClass to store the information in a XMS matrix table. The record inherits from a list containing the individual motifs. cCsddd|DS)zBReturn a string representation of the motifs in the Record object. css|]}t|VqdS)N)str)rrrrr ^sz!Record.__str__..)rG)r rrr__str__\szRecord.__str__N)rJrKrLrMrQrrrrrVs rcCst|}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)handlexms_docrrrrreadas  rU) rMrExml.domrrBiorrlistrrUrrrrs   F