U Qbi@sNddlmZddlZddlZedddddddZd ed <Gd d d eZdS) ) defaultdictNcCsdS)Nrrr.lib/python3.8/site-packages/deblur/sequence.pyr)ACGT-c@s0eZdZdZddZddZddZdd Zd S) Sequencea Sequence object to represent the aligned reads Attributes ---------- label : str The sequence label sequence : str The sequence string length : int The sequence length (aligned) unaligned_length : int The unaligned sequence length frequency : float The number of times the sequence have been seen in the dataset np_sequence : numpy array of int8 An int8 numpy representation of the sequence string Methods ------- to_fasta cCsl||_||_t|j|_|j|jd|_tt d|j d|_ t j dd|jDt jd|_dS)Nr (?<=size=)\w+rcSsg|] }t|qSr) trans_dict).0brrr 1sz%Sequence.__init__..)Zdtype)labeluppersequencelenlengthcountZunaligned_lengthfloatresearchgroup frequencynpZarrayZint8Z np_sequence)selfrrrrr__init__)s  zSequence.__init__cCs(t|t|ko&|j|jko&|j|jkSN)typerr!r#otherrrr__eq__3s   zSequence.__eq__cCs || Sr%)r)r'rrr__ne__8szSequence.__ne__cCs@tjd|jdd\}}tt|j}d|||f}d||jfS)zReturns a string with the sequence in fasta format Returns ------- str The FASTA representation of the sequence rr)maxsplitz%s%d%sz>%s %s )rsplitrintroundr!r)r#prefixsuffixZ new_countZ new_labelrrrto_fasta;szSequence.to_fastaN)__name__ __module__ __qualname____doc__r$r)r*r1rrrrrs  r) collectionsrrZnumpyr"robjectrrrrr s