3 y^)@sddlmZmZmZmZddlZddlZddlZddlZddl Z ddl Z ddl Z ddl Z ddl Z ddlZejeZGdddZdS))absolute_importdivisionprint_functionunicode_literalsNc@s<eZdZdZddZddZddZdd Zed d Z d S) Node_pathz Object representation of the connected set of Node objects that represent the reconstructed isoforms graph traversal Instance members: transcript_name : (str) name of the isoform node_obj_list : (list) of Node objects cCs||_t|_tjd|}d}t}xn|D]f}|jd\} } | jd\} } t| } t| } |sjd}| d7} |j|| || | d} |jj| q*WdS) zV constructor, instantiates Node_path and builds vertices in the graph z \d+:\d+\-\d+F:-TfstN) transcript_namelist node_obj_listrefindallsplitintget_nodeappend)selftgraphr Z path_stringsequenceZnode_descr_listZfirst_kmer_flagZ obj_node_listZ node_descr loc_node_idZnode_coord_rangeZlendZrendnode_objrM/broad/hptmp/bhaas/trinityrnaseq/Analysis/SuperTranscripts/pylib/Node_path.py__init__!s   zNode_path.__init__cCs|jS)N)r )rrrrget_transcript_name?szNode_path.get_transcript_namecCs|jS)N)r )rrrrget_pathBszNode_path.get_pathcCs2t}x|jD]}|jt|qWdj|}|S)Nz--)r r rstrjoin)rZ node_str_listnodepath_strrrr__repr__Fs   zNode_path.__repr__cCst}|j}xN|D]F}|j}tjd|rtjdd|}|j|}|dk r|j||fqW|sptj d|Stj dj |t }t} x|D]\} }| j } |j } | ddd } | ddd}tj || stdj | | t| t| }|dkr2| | kstd |j| j|g|| <| j| q| d|}tj d j | | ||j| j| |g|| <| j|qWx|D]x}|j}|d}||kr||}t|dkr|d|d<n4t|d kr|d|d<|jd|dntd qWx| D]}|j|qW|S)aX fst nodes will have an extra 5' sequence as compared to the corresponding non-fst nodes. If both the fst and non-fst version of the node exist, must modify the fst nodes so that they are separated from their 5' extension, and the core of the node (suffix) is shared. input: TGraph obj, list of node_path objects. The node_path objects are modified in-place as needed. A fst-node will be truncated to the unique prefix and the non-fst node will be integrated into the path. returns the node_path_list with any required adjustments r Nzno FST nodes to adjustzAdjusting FST nodes: {}r z8Error, core_node_seq: {} is not a suffix of fst seq: {} rzGError, prefix starts at first position but sequences are not equivalentzs