:>"^@sdZddlmZddlmZddlmZddlmZddl m Z eddZ e d krdd l mZed dd S) a<Bio.SeqIO support for the "ig" (IntelliGenetics or MASE) file format. This module is for reading and writing IntelliGenetics format files as SeqRecord objects. This file format appears to be the same as the MASE multiple sequence alignment format. You are expected to use this module via the Bio.SeqIO functions. )print_function) as_handle)single_letter_alphabet)Seq) SeqRecordc cst|d}x'|j}|s(P|jdsPqWxM|r|ddkretd|g}x<|jdr|j|ddj|j}qnW|j}g}xG|j}|sP|ddkrP|j|jjdd qWd j|}|j d r4|dd}d |krLtd t t ||d |d |}dj||j d<|Vq?W| st WdQRXdS)a/Iterate over IntelliGenetics records (as SeqRecord objects). handle - input file alphabet - optional alphabet The optional free format file header lines (which start with two semi-colons) are ignored. The free format commentary lines at the start of each record (which start with a semi-colon) are recorded as a single string with embedded new line characters in the SeqRecord's annotations dictionary under the key 'comment'. Examples -------- >>> with open("IntelliGenetics/TAT_mase_nuc.txt") as handle: ... for record in IgIterator(handle): ... print("%s length %i" % (record.id, len(record))) ... A_U455 length 303 B_HXB2R length 306 C_UG268A length 267 D_ELI length 309 F_BZ163A length 309 O_ANT70 length 342 O_MVP5180 length 348 CPZGAB length 309 CPZANT length 309 A_ROD length 390 B_EHOA length 420 D_MM251 length 390 STM_STM length 387 VER_AGM3 length 354 GRI_AGM677 length 264 SAB_SAB1C length 219 SYK_SYK length 330 rUz;;r;z)Records should start with ';' and not: %rN 1z5Potential terminator digit one found within sequence.idname comment)rreadline startswith ValueErrorappendstriprstripreplacejoinendswithrr annotationsAssertionError)handleZalphabetlineZ comment_linestitleZ seq_linesZseq_strrecordr!3/tmp/pip-build-ww9dw3qa/biopython/Bio/SeqIO/IgIO.py IgIteratorsB'    #  ! r#__main__) run_doctestverboseN)__doc__ __future__rZBio.FilerZ Bio.AlphabetrZBio.SeqrZ Bio.SeqRecordrr#__name__Z Bio._utilsr%r!r!r!r"s]