o Rŀgk@sdZddlZddlmmZddlmZddlm Z GdddZ GdddZ Gd d d Z Gd d d Z Gd ddZGdddZdS)a'Classes to represent a KGML Pathway Map. The KGML definition is as of release KGML v0.7.2 (http://www.kegg.jp/kegg/xml/docs/) Classes: - Pathway - Specifies graph information for the pathway map - Relation - Specifies a relationship between two proteins or KOs, or protein and compound. There is an implied direction to the relationship in some cases. - Reaction - A specific chemical reaction between a substrate and a product. - Entry - A node in the pathway graph - Graphics - Entry subelement describing its visual representation N)chain)minidomc@seZdZdZddZddZddZdd Zd d Zd d Z ddZ ddZ ddZ ddZ ddZddZee eedZddZddZdd Zeeeed!Zed"d#Zed$d%Zed&d'Zed(d)Zed*d+Zed,d-Zed.d/Zed0d1Zed2d3Zd4S)5PathwayaRepresents a KGML pathway from KEGG. Specifies graph information for the pathway map, as described in release KGML v0.7.2 (http://www.kegg.jp/kegg/xml/docs/) Attributes: - name - KEGGID of the pathway map - org - ko/ec/[org prefix] - number - map number (integer) - title - the map title - image - URL of the image map for the pathway - link - URL of information about the pathway - entries - Dictionary of entries in the pathway, keyed by node ID - reactions - Set of reactions in the pathway The name attribute has a restricted format, so we make it a property and enforce the formatting. The Pathway object is the only allowed route for adding/removing Entry, Reaction, or Relation elements. Entries are held in a dictionary and keyed by the node ID for the pathway graph - this allows for ready access via the Reaction/Relation etc. elements. Entries must be added before reference by any other element. Reactions are held in a dictionary, keyed by node ID for the path. The elements referred to in the reaction must be added before the reaction itself. cCs<d|_d|_d|_d|_d|_d|_i|_i|_t|_ dSInitialize the class.N) _nameorg_numbertitleimagelinkentries _reactionsset _relationsselfrN/var/www/html/myenv/lib/python3.10/site-packages/Bio/KEGG/KGML/KGML_pathway.py__init__As zPathway.__init__cCsHddddtdg}|t|jd}t|}|j ddS) z9Return the pathway as a string in prettified KGML format. zzJz zutf-8z )indent) jointimeasctimeETtostringelementdecoder parseString toprettyxml)rheader rough_xmlreparsedrrrget_KGMLMs  zPathway.get_KGMLcCs@t|jtstdt|jd|jd||_||j|j<dS)z$Add an Entry element to the pathway. Node ID must be an integer, got  ()N) isinstanceidint TypeErrortype_pathwayrrentryrrr add_entry\s zPathway.add_entrycC8t|jtstdt|jd|jd|j|j=dS)z)Remove an Entry element from the pathway.r&r'r(N)r)r*r+r,r-rr/rrr remove_entryf zPathway.remove_entrycCsZt|jtstdt|jd|jd|j|jvr"td|j||_||j|j<dS)z&Add a Reaction element to the pathway.r&r'r(z)Reaction ID %d has no corresponding entryN)r)r*r+ ValueErrorr-rr.rrreactionrrr add_reactionqs  zPathway.add_reactioncCr2)z+Remove a Reaction element from the pathway.r&r'r(N)r)r*r+r,r-rr6rrrremove_reaction}r4zPathway.remove_reactioncCs||_|j|dS)z&Add a Relation element to the pathway.N)r.raddrrelationrrr add_relationszPathway.add_relationcC|j|dS)z+Remove a Relation element from the pathway.N)rremover;rrrremove_relationzPathway.remove_relationcsd|jd|jd|jd|jdt|jdg}dD]fdd |jD}t|r<|d t|fqd |d S) z-Return a readable summary description string.z Pathway: z KEGG ID: z Image file: z Organism: z Entries: %dz Entry types:)orthologenzymer7genegroupcompoundmapcsg|] }|jkr|qSrr-.0etrr z#Pathway.__str__..z %s: %dr) r namer r lenrvaluesappendr)routstretyperrLr__str__s     zPathway.__str__cC|jSNrrrrr_getnamezPathway._getnamecCs"|ds td|||_dS)Nzpath:z,Pathway name should begin with 'path:', got ) startswithr5rrvaluerrr_setnames  zPathway._setnamecC|`dSrXrYrrrr_delnamezPathway._delnamezThe KEGGID for the pathway map.cCrWrXr rrrr _getnumberr[zPathway._getnumbercCt||_dSrX)r+r r]rrr _setnumberzPathway._setnumbercCr`rXrcrrrr _delnumberrbzPathway._delnumberzThe KEGG map number.cCdd|jDS)z'Get a list of entries of type compound.cSg|] }|jdkr|qS)rFrHrIrrrrNrOz%Pathway.compounds..rrRrrrr compoundszPathway.compoundscCri)z"Get a list of entries of type map.cSrj)rGrHrIrrrrNrOz Pathway.maps..rkrrrrmapsrmz Pathway.mapscCri)z'Get a list of entries of type ortholog.cSrj)rBrHrIrrrrNrOz%Pathway.orthologs..rkrrrr orthologsrmzPathway.orthologscCri)z#Get a list of entries of type gene.cSrj)rDrHrIrrrrNrOz!Pathway.genes..rkrrrrgenesrmz Pathway.genescCs |jS)z'Get a list of reactions in the pathway.)rrRrrrr reactions zPathway.reactionscfddjDS)z>List of entries corresponding to each reaction in the pathway.csg|]}j|qSr)rrJirrrrNsz,Pathway.reaction_entries..rrrrrreaction_entriesrmzPathway.reaction_entriescCs t|jS)z'Get a list of relations in the pathway.)listrrrrr relationsrrzPathway.relationscCstd}|j|jt|j|j|j|jd|_ t |j D] \}}| |jq|jD]}| |jq,t |j D] \}}| |jq<|S)z+Return the Pathway as a valid KGML element.pathway)rPr numberr r r )rElementrr strr r r r attribsortedritemsrSrrr)rrzeidr0r<r7rrrrs   zPathway.elementcCsgg}}dd|jDD] }||dd|ddg||dd|ddgqt|t|ft|t|fgS)zCoordinate bounds for all Graphics elements in the Pathway. Returns the [(xmin, ymin), (xmax, ymax)] coordinates for all Graphics elements in the Pathway cSg|]}|jqSrboundsrJgrrrrNz"Pathway.bounds..r)rrRextendminmaxrxlistylistbrrrrs   zPathway.boundsN)__name__ __module__ __qualname____doc__rr%r1r3r8r9r=r@rVrZr_rapropertyrPrdrfrhr{rlrnrorprqrwryrrrrrrr sJ              rc@seZdZdZddZddZddZdd Zd d Zd d Z ddZ ddZ ddZ e e e e dZddZddZddZe eeedZddZddZd d!Ze eeed"Ze d#d$Ze d%d&Ze d'd(Zd)S)*EntryaRepresent an Entry from KGML. Each Entry element is a node in the pathway graph, as described in release KGML v0.7.2 (http://www.kegg.jp/kegg/xml/docs/) Attributes: - id - The ID of the entry in the pathway map (integer) - names - List of KEGG IDs for the entry - type - The type of the entry - link - URL of information about the entry - reaction - List of KEGG IDs of the corresponding reactions (integer) - graphics - List of Graphics objects describing the Entry's visual representation - components - List of component node ID for this Entry ('group') - alt - List of alternate names for the Entry NOTE: The alt attribute represents a subelement of the substrate and product elements in the KGML file cCsBd|_g|_d|_d|_d|_g|_t|_g|_d|_ g|_ dSrNr) _id_namesr-r r graphicsr componentsaltr.rrrrrrs zEntry.__init__cCsTd|jd|jd|jd|jd|jdt|j|jfg}d|dS)z#Return readable descriptive string.zEntry node ID: %dzNames: Type: z Components: z Reactions: zGraphics elements: %d %sr)r*rPr-rr7rQrrrrTrrrrV's    z Entry.__str__cCs:|jdur|j|jjvrtd|jd|j|dS)zAdd an element to the entry. If the Entry is already part of a pathway, make sure the component already exists. Nz Component z is not an entry in the pathway)r.r*rr5rr:)rrrrr add_component3s  zEntry.add_componentcCr>)z3Remove the entry with the passed ID from the group.N)rr?r]rrrremove_component@rAzEntry.remove_componentcCr>)zAdd the Graphics entry.N)rrSr/rrr add_graphicsDrAzEntry.add_graphicscCr>)z.rr)rrrrrrrrrs   z Entry.boundscCs$|jjD] }|j|jvrdSqdS)zMReturn true if this Entry participates in any reaction in its parent pathway.TF)r.rqr reactant_ids)rrxnrrr is_reactants  zEntry.is_reactantN)rrrrrrVrrrrrZr_rarrPrrrr7rrrr*rrrrrrrrsB      rc@sJeZdZdZddZddZddZdd Zeeeed Z ed d Z d S) ComponentaYAn Entry subelement used to represents a complex node. A subelement of the Entry element, used when the Entry is a complex node, as described in release KGML v0.7.2 (http://www.kegg.jp/kegg/xml/docs/) The Component acts as a collection (with type 'group', and typically its own Graphics subelement), having only an ID. cCsd|_||_dS)rN)r_parentrparentrrrrs zComponent.__init__cCrWrXrrrrrrr[zComponent._getidcCrerXrr]rrrrrgzComponent._setidcCr`rXrrrrrrrbzComponent._delidz'The pathway graph node ID for the EntrycCstd}dt|ji|_|S)z-Return the Component as a valid KGML element. componentr*)rr|r}rr~)rrrrrrs zComponent.elementN) rrrrrrrrrr*rrrrrrs rc@sFeZdZdZddZddZddZdd Zeeeed Z d d Z d dZ ddZ ee e e dZ ddZddZddZeeeedZddZddZddZeeeedZd d!Zd"d#Zd$d%Zeeeed&Zd'd(Zd)d*Zd+d,Zeeeed-Zd.d/Zd0d1Zd2d3Z eeee d4Z!ed5d6Z"ed7d8Z#ed9d:Z$d;S)<GraphicsaAn Entry subelement used to represents the visual representation. A subelement of Entry, specifying its visual representation, as described in release KGML v0.7.2 (http://www.kegg.jp/kegg/xml/docs/) Attributes: - name Label for the graphics object - x X-axis position of the object (int) - y Y-axis position of the object (int) - coords polyline coordinates, list of (int, int) tuples - type object shape - width object width (int) - height object height (int) - fgcolor object foreground color (hex RGB) - bgcolor object background color (hex RGB) Some attributes are present only for specific graphics types. For example, line types do not (typically) have a width. We permit non-DTD attributes and attribute settings, such as dash List of ints, describing an on/off pattern for dashes cCs@d|_d|_d|_d|_d|_d|_d|_d|_d|_||_ dSr) rP_x_y_coordsr-_width_heightfgcolorbgcolorrrrrrrs zGraphics.__init__cCrWrXrrrrr_getxr[zGraphics._getxcCrerX)floatrr]rrr_setxrgzGraphics._setxcCr`rXrrrrr_delxrbzGraphics._delxz*The X coordinate for the graphics element.cCrWrXrrrrr_getyr[zGraphics._getycCrerX)rrr]rrr_setyrgzGraphics._setycCr`rXrrrrr_delyrbzGraphics._delyz*The Y coordinate for the graphics element.cCrWrXrrrrr _getwidthr[zGraphics._getwidthcCrerX)rrr]rrr _setwidth rgzGraphics._setwidthcCr`rXrrrrr _delwidth rbzGraphics._delwidthz"The width of the graphics element.cCrWrXrrrrr _getheightr[zGraphics._getheightcCrerX)rrr]rrr _setheightrgzGraphics._setheightcCr`rXrrrrr _delheightrbzGraphics._delheightz#The height of the graphics element.cCrWrXrrrrr _getcoords!r[zGraphics._getcoordscs8dd|dDfddtdtdD|_dS)NcSg|]}t|qSr)r+rIrrrrN%z'Graphics._setcoords..,cs g|] }t||dqS))tuplertclistrrrN&s rr)rrangerQrr]rrr _setcoords$s$zGraphics._setcoordscCr`rXrrrrr _delcoords(rbzGraphics._delcoordsz.Polyline coordinates for the graphics element.cCrWrX_fgcolorrrrr _getfgcolor3r[zGraphics._getfgcolorcC|dkr d|_dS||_dSNnonez#000000rr]rrr _setfgcolor6  zGraphics._setfgcolorcCr`rXrrrrr _delfgcolor<rbzGraphics._delfgcolorzForeground color.cCrWrX_bgcolorrrrr _getbgcolorAr[zGraphics._getbgcolorcCrrrr]rrr _setbgcolorDrzGraphics._setbgcolorcCr`rXrrrrr _delbgcolorJrbzGraphics._delbgcolorzBackground color.cCstd}t|jtr|j}n d|jdd}t|jtr$|j}n d|jdd}|j|j||d|_ dD]\}}t ||durPtt |||j |<q;|jdkrgd d d t |jD|j d <|S) z,Return the Graphics as a valid KGML element.r#rN)rPr-rr))xr)yr)widthr)heightrlinercSrr)r}rIrrrrNlrz$Graphics.element..coords)rr|r)rr}hexvalrrPr-r~getattrrr from_iterabler)rrfghexbghexnattrrrrrOs*      zGraphics.elementcCs|jdkr%dd|jD}dd|jD}t|t|ft|t|fgS|j|jd|j|jdf|j|jd|j|jdfgS)a,Coordinate bounds for the Graphics element. Return the bounds of the Graphics object as an [(xmin, ymin), (xmax, ymax)] tuple. Coordinates give the centre of the circle, rectangle, roundrectangle elements, so we have to adjust for the relevant width/height. rcSsg|]\}}|qSrrrJrrrrrrNzrz#Graphics.bounds..cSsg|]\}}|qSrrrrrrrN{r?)r-rrrrrrr)rrrrrrrps  zGraphics.boundscCs@d|jdd|jddd|jdd|jddfS)z.z Products: cSrrrrrrrrNrr)r*rPr-r substratesproductsrrrrrVs   zReaction.__str__cCs:|jdurt||jjvrtdt||j|dS)z.)rrrrrrrmzReaction.substratescrs)z&Return list of product Entry elements.crrr)rJpidrrrrNrz%Reaction.products..)rrrrrrrmzReaction.productscCs|jj|jS)z0Return the Entry corresponding to this reaction.)r.rrrrrrr0zReaction.entrycCs|j|jS)z4Return a list of substrate and product reactant IDs.)runionrrrrrrrzReaction.reactant_idscCstd}t|j|j|jd|_|jD]}td}t||jd<|jj |j|jd<| |q|j D]}td}t||jd<|jj |j|jd<| |q5|S)z,Return KGML element describing the Reaction.r7)r*rPr- substrater*rPproduct) rr|r}r*rPr-r~rr.rrSr)rr7rrprrrrrs       zReaction.elementN)rrrrrrVr r rrrrr*r r rrPrrr0rrrrrrrs4        rc@sxeZdZdZddZddZddZdd Zd d Ze eeed Z d dZ ddZ ddZ e e e e dZe ddZdS)Relationa"A relationship between to products, KOs, or protein and compound. This describes a relationship between two products, KOs, or protein and compound, as described in release KGML v0.7.2 (http://www.kegg.jp/kegg/xml/docs/) Attributes: - entry1 - The first Entry object node ID defining the relation (int) - entry2 - The second Entry object node ID defining the relation (int) - type - The relation type - subtypes - List of subtypes for the relation, as a list of (name, value) tuples cCs"d|_d|_d|_g|_d|_dSr)_entry1_entry2r-subtypesr.rrrrrs  zRelation.__init__cCsZdt|jdt|jdt|jg}|jD]}|d|dt|dgqd|S)z&Return a useful human-readable string.zRelation (subtypes: %d):zEntry1:zEntry2:z Subtype: rrr)rQrr}entry1entry2rr)rrTrrrrrV#s  " zRelation.__str__cC|jdur |jj|jS|jSrX)r.rrrrrr _getentry11 zRelation._getentry1cCrerX)r+rr]rrr _setentry16rgzRelation._setentry1cCr`rX)rrrrr _delentry19rbzRelation._delentry1zEntry1 of the relation.cCr rX)r.rrrrrr _getentry2>r"zRelation._getentry2cCrerX)r+rr]rrr _setentry2CrgzRelation._setentry2cCr`rX)rrrrr _delentry2FrbzRelation._delentry2zEntry2 of the relation.cCs^td}t|jt|j|jd|_|jD]\}}td}|t|d|_||q|S)z,Return KGML element describing the Relation.r<)rrr-subtype)rPr^) rr|r}rrr-r~rrS)rr<rPr^r(rrrrKs   zRelation.elementN)rrrrrrVr!r#r$rrr%r&r'rrrrrrr sr)rrxml.etree.ElementTreeetree ElementTreer itertoolsrxml.domrrrrrrrrrrrs  d!&E|