RidZddlZddlZddlZddlmZddlmZddlm Z ddlm Z Gdd e Z Gd d e Z e d k(rdd lmZedyy)aBio.SeqIO support for the UCSC nib file format. Nib stands for nibble (4 bit) representation of nucleotide sequences. The two nibbles in a byte each store one nucleotide, represented numerically as follows: - ``0`` - T - ``1`` - C - ``2`` - A - ``3`` - G - ``4`` - N (unknown) As the first bit in a nibble is set if the nucleotide is soft-masked, we additionally have: - ``8`` - t - ``9`` - c - ``a`` - a - ``b`` - g - ``c`` - n (unknown) A nib file contains only one sequence record. You are expected to use this module via the Bio.SeqIO functions under the format name "nib": >>> from Bio import SeqIO >>> record = SeqIO.read("Nib/test_even_bigendian.nib", "nib") >>> print("%i %s..." % (len(record), record.seq[:20])) 50 nAGAAGagccgcNGgCActt... For detailed information on the file format, please see the UCSC description at https://genome.ucsc.edu/FAQ/FAQformat.html. N)Seq) SeqRecord)SequenceIterator)SequenceWriterc,eZdZdZdZfdZdZxZS) NibIteratorzParser for nib files.bct||d|jjd}|s t d|j }|dk(rd|_y |dk(rd|_y t d ) aIterate over a nib file and yield a SeqRecord. - source - a file-like object or a path to a file in the nib file format as defined by UCSC; the file must be opened in binary mode. Note that a nib file always contains only one sequence record. The sequence of the resulting SeqRecord object should match the sequence generated by Jim Kent's nibFrag utility run with the -masked option. This function is used internally via the Bio.SeqIO functions: >>> from Bio import SeqIO >>> record = SeqIO.read("Nib/test_even_bigendian.nib", "nib") >>> print("%s %i" % (record.seq, len(record))) nAGAAGagccgcNGgCActtGAnTAtCGTCgcCacCaGncGncTtGNtGG 50 You can also call it directly: >>> with open("Nib/test_even_bigendian.nib", "rb") as handle: ... for record in NibIterator(handle): ... print("%s %i" % (record.seq, len(record))) ... nAGAAGagccgcNGgCActtGAnTAtCGTCgcCacCaGncGncTtGNtGG 50 Nib)fmtz Empty file.3a3de96blittle6be93d3abigz"unexpected signature in nib headerN)super__init__streamread ValueErrorhex byteorder)selfsourceword signature __class__s G/home/agent/.friday_env/lib/python3.12/site-packages/Bio/SeqIO/NibIO.pyrzNibIterator.__init__:sn4 U+{{"]+ +HHJ  "%DN * $"DNAB Bc6|j}|j}|jd}|sttj ||}|j}t j|}|dzdk(rt||k7r4td|dzdk(r!t||dzk7r td|d|}t|jds tdtjdd }|j|}t|} t!| } | S) z)Iterate over the records in the nib file.rrzUnexpected file sizerN 0123489abcz&Unexpected sequence data found in file TCAGNtcagn)rrr StopIterationint from_bytesbinasciihexlifylenrsetissubsetbytes maketrans translaterr) rrrnumberlengthdataindicestable nucleotidessequencerecords r__next__zNibIterator.__next__`sNN Q  2{{}""4( A:?7|v% !788 aZ1_7|vz) !788gv&G7|$$]3EF F }=''. {#8$ r )__name__ __module__ __qualname____doc__modesrr8 __classcell__)rs@rr r 5s E$CLr r c&eZdZdZdZdZdZdZy) NibWriterzNib file writer.r c|j}tj}|dk(rd}n|dk(rd}ntd||j t j |y)zWrite the file header.rrrrzunexpected system byte order N)handlesysr RuntimeErrorwriter-fromhex)rrBrrs r write_headerzNibWriter.write_headersVMM  "I % "I!>ykJK K U]]9-.r c|j}|j}t|}t|}|j t j d|tjdd}|dz}|dz}||z }t|jds td|j|} |j tj| y) z)Write a single record to the output file.ir$r#r"Ts ACGTNacgtnz0Sequence should contain A,C,G,T,N,a,c,g,t,n onlyN)rBseqr-r*rEstructpackr.r+r,rr/r( unhexlify) rr7rBr6r5r1r4paddingsuffixr3s r write_recordzNibWriter.write_records::Ho X V[[f-. }=1*4v ;((7OP P''. X''01r c|jd}|D](}|dk(r td|j||dz }*|dk7r td|S)zWrite records to the output file, and return the number of records. records - A list or iterator returning SeqRecord objects rrzMore than one sequence foundzMust have one sequence)rGrrQ)rrecordscountr7s r write_recordszNibWriter.write_recordssh  Fz !?@@   f % QJE   A:56 6 r N)r9r:r;r<r=rGrQrUr rr@r@zs E /2 r r@__main__) run_doctest)verbose)r<r(rLrCBio.Seqr Bio.SeqRecordr Interfacesrrr r@r9 Bio._utilsrXrVr rr^sY D #(&B"BJ//d z&r