U _v@sFddlZddlZddlZddlmZmZddlmZGdddZdS)N)strftime localtime)versionc@sJeZdZdZdddZddZdd Zd d Zd d ZddZ ddZ dS) GFFWritera Simple GFF writer class for serializing gffutils records to a file. Parameters: ----------- out: string or file-like object If a string, parsed as a filename, otherwise, a file-like object to write to. with_header: bool If True, output a header file for the GFF. The header indicates the file was created with gffutils and includes a timestamp and version info. in_place: bool If True and if `out` is a filename, then write the file in place (uses named temporary files.) TODO: Add make separate GTFWriter class or add support for GTF here. TFcCs||_||_||_d|_d|_t|tjr`|jrPtj dd|_t |jj d|_qtt |jd|_n|jrnt d||_|jrt dt}dt|f}|jd|dS)NF)deletewz/Cannot use 'in_place' when writing to a stream.z%Y-%m-%d %H:%M:%Sz,#GFF3 file (created by gffutils (v%s) on %s)%s )out with_headerin_place temp_file out_stream isinstancesixZ string_typestempfileZNamedTemporaryFileopenname ValueErrorrrrwrite)selfr r r Z timestampheaderr1lib/python3.8/site-packages/gffutils/gffwriter.py__init__#s&  zGFFWriter.__init__cCs|jd|dS)z( Output record to file. rN)r r)rrecrrr write_recBszGFFWriter.write_reccCs|D]}||qdS)z1 Output several records to file. N)r)rZrecsrrrr write_recsHszGFFWriter.write_recsc Cs||}||i}t|j|dd}|j|ddD]&}tdd|j|ddD||j<q6t|dddd }|D]*}|d } || } || ||| qx|j|d d D]} | jdkr|| qd S)a NOTE: The goal of this function is to have a canonical ordering when outputting a gene and all of its records to a file. The order is intended to be: gene # mRNAs sorted by length, with longest mRNA first mRNA_1 # Exons of mRNA, sorted by start position (ascending) exon_1 # Children of exon, sorted by start position exon_child_1 exon_child_2 exon_2 ... # Non-exonic children here ... mRNA_2 ... # Non-mRNA children here ... Output records of a gene to a file, given a GFF database and a gene_id. Outputs records in canonical order: gene record first, then longest mRNA, followed by longest mRNA exons, followed by rest, followed by next longest mRNA, and so on. Includes the gene record itself in the output. TODO: This probably doesn't handle deep GFF hierarchies. mRNA) featuretypecss|]}t|VqdS)N)len).0exonrrr wsz,GFFWriter.write_gene_recs..r!cSs|dS)Nr)xrrr{z+GFFWriter.write_gene_recs..T)keyreverserr#)levelN) rlistchildrensumidsorteditemswrite_mRNA_childrenr) rdbZgene_idZgene_recZ mRNA_lenscrZ sorted_mRNAsZ curr_mRNAmRNA_idZmRNA_recZ gene_childrrrwrite_gene_recsOs&     zGFFWriter.write_gene_recscCsV|j|dd}g}|D]0}|jdkr<|||||q||q||dS)z Write out the children records of the mRNA given by the ID (not including the mRNA record itself) in a canonical order, where exons are sorted by start position and given first. startZorder_byr!N)r+rrwrite_exon_childrenappendr)rr1r3Z mRNA_childrenZnonexonic_childrenZ child_recrrrr0s   zGFFWriter.write_mRNA_childrencCs&|j|dd}|D]}||qdS)z| Write out the children records of the exon given by the ID (not including the exon record itself). r5r6N)r+r)rr1Zexon_idZ exon_childrenZ exon_childrrrr7szGFFWriter.write_exon_childrencCs&|j|jr"t|jj|jdS)zF Close the stream. Assumes stream has 'close' method. N)r closer shutilmover rr )rrrrr9s zGFFWriter.closeN)TF) __name__ __module__ __qualname____doc__rrrr4r0r7r9rrrrr s 9 r) rrr:timerrZgffutils.versionrrrrrrs