o Rŀg&@sldZddlZddlZddlmZddlmZddlmZddl m Z Gdddej Z Gd d d ej Z dS) azBio.Align support for BED (Browser Extensible Data) files. The Browser Extensible Data (BED) format, stores a series of pairwise alignments in a single file. Typically they are used for transcript to genome alignments. BED files store the alignment positions and alignment scores, but not the aligned sequences. See http://genome.ucsc.edu/FAQ/FAQformat.html#format1 You are expected to use this module via the Bio.Align functions. Coordinates in the BED format are defined in terms of zero-based start positions (like Python) and aligning region sizes. A minimal aligned region of length one and starting at first position in the source sequence would have ``start == 0`` and ``size == 1``. As we can see in this example, ``start + size`` will give one more than the zero-based end position. We can therefore manipulate ``start`` and ``start + size`` as python list slice boundaries. N) Alignment) interfaces)Seq) SeqRecordcs*eZdZdZdfdd ZddZZS)AlignmentWriterzHAlignment file writer for the Browser Extensible Data (BED) file format. cs.|dks|dkr tdt|||_dS)zCreate an AlignmentWriter object. Arguments: - target - output stream or file name - bedN - number of columns in the BED file. This must be between 3 and 12; default value is 12. rzbedN must be between 3 and 12N) ValueErrorsuper__init__bedN)selftargetr  __class__A/var/www/html/myenv/lib/python3.10/site-packages/Bio/Align/bed.pyr *s   zAlignmentWriter.__init__cCst|ts td|j}|jsdS|j}|j\}}z|j}Wn ty)d}Ynw|dur0d}|d|dkrC|dddddf}|d|d krNd }nd }g}g} |ddd f\} } |ddd df D]$\} } | | kru| } qj| | kr|| } qj| | }| | | || } | } qjz | d }| d|}Wn t y| }}Ynw|t |t |g}|dkrd |dSz|j}Wn tyd}Ynw|durd}| ||dkrd |dSz|j}Wn tyd }Ynw| t|d|dkr d |dS| ||dkrd |dSz|j}Wn ty+|}Ynw| t ||dkr?d |dSz|j}Wn tyP|}Ynw| t ||dkrdd |dSz|j}Wn tyud}Ynw| t ||dkrd |dSt|}| t ||dkrd |dS| d tt |d|dkrd |dS| |8} | d tt | dd |dS)z;Return a string with one alignment formatted as a BED line.zExpected an Alignment objectrNrrrr)r)rr-+rrr  queryg0 , ) isinstancer TypeError coordinatessizer sequencesidAttributeError transposeappend IndexErrorstrjoinscoreformat thickStartthickEnditemRgblenmap)r alignmentr*r rrchromstrand blockSizes blockStartstStartqStarttEndqEnd blockSize chromStartchromEndfieldsnamer4r6r7r8 blockCountrrrformat_alignment8s                         z AlignmentWriter.format_alignment)r)__name__ __module__ __qualname____doc__r rJ __classcell__rrrrr'src@seZdZdZdZddZdS)AlignmentIteratorzAlignment iterator for Browser Extensible Data (BED) files. Each line in the file contains one pairwise alignment, which are loaded and returned incrementally. Additional alignment information is stored as attributes of each alignment. BEDc Cs6|D]}|}t|}|dks|dkrtd||d}t|d}t|d}|dkr4|d}nd}|dkr?|d} nd} |d krt|d } d d |d d d D} dd |dd d D} t| | krytdt| | ft| | krtdt| | ft| } t| } d} d}| |gg}t| | D] \}}|| kr|||g|} | |7} ||7}|| |gqt| }t | }n||}td|gd|gg}|}|dddf|7<t d|d}t ||dd}t dt jd}t ||dd}||g}| dkr||dddf|dddf<||dkr0td||df||dkrAtd||dft||}|dkrO|S|d}zt|}Wn tycYnw||_|dkrp|St|d|_|dkr|St|d|_|dkr|S|d|_|SdS)Nrrz+expected between 3 and 12 columns, found %drrrrr$cSg|]}t|qSrint).0rDrrr z:AlignmentIterator._read_next_alignment..r%r&cSrSrrT)rV blockStartrrrrWrXr'z:Inconsistent number of block sizes (%d found, expected %d)zDInconsistent number of block start positions (%d found, expected %d))lengthr)r- descriptionrrz/Inconsistent chromStart found (%d, expected %d)rz-Inconsistent chromEnd found (%d, expected %d)rr r!r")splitr9r rUrstripnparrayzipr0r/sumrrsysmaxsizerfloatr4r6r7r8)r streamlinewordsr r<rErFrHr=rIr>r? tPosition qPositionr*rDrYqSizequery_sequence query_recordtarget_sequence target_recordrecordsr;r4rrr_read_next_alignments                           z&AlignmentIterator._read_next_alignmentN)rKrLrMrNfmtrprrrrrPs rP) rNrbnumpyr^ Bio.AlignrrBio.Seqr Bio.SeqRecordrrrPrrrrs    v