U QÁ™bþã@sFddlmZddlZddlZddlmZdd„Zdd„Zd d d „Z dS)é)Ú attrgetterN©ÚSequencec Cs ddddddddddddg }|S)zSReturn the default error profile for deblurring based on illumina run data ég¸…ëQ¸®?g{®Gáz”?ç{®Gáz„?ç{®Gázt?gü©ñÒMbP?gü©ñÒMb@?©)Ú error_distrrú0lib/python3.8/site-packages/deblur/deblurring.pyÚget_default_error_profiles þr cCsÊzdd„|Dƒ}Wntk r*g}YnXt|ƒdkrPt t¡}| d¡dStdd„|Dƒƒ}tdd„|Dƒƒ}t|ƒd ksŒt|ƒd kr´td d  t t |ƒ¡d  t t |ƒ¡fƒ‚t |t d ƒd d}|S)aReturns a list of Sequences Parameters ---------- input_seqs : iterable of (str, str) The list of input sequences in (label, sequence) format Returns ------- list of Sequence Raises ------ ValueError If no sequences where found in `input_seqs` If all the sequences do not have the same length either aligned or unaligned. cSsg|]\}}t||ƒ‘qSrr)Ú.0ÚidÚseqrrr Ú /sz!get_sequences..rz!No sequences found in fasta file!Ncss|] }|jVqdS©N)Úlength©r Úsrrr Ú 9sz get_sequences..css|] }|jVqdSr)Úunaligned_lengthrrrr r:srzPNot all sequence have the same length. Aligned lengths: %s, sequence lengths: %sz, Ú frequencyT)ÚkeyÚreverse) Ú ExceptionÚlenÚloggingÚ getLoggerÚ__name__ÚwarnÚsetÚ ValueErrorÚjoinÚmapÚstrÚsortedr)Ú input_seqsÚseqsÚloggerZaligned_lengthsZunaligned_lengthsrrr Ú get_sequencess(    ÿþÿr(rréc Csêt t¡}|dkrtƒ}| d|¡t|ƒ}|dkrD| d¡dS| dt|ƒ¡t d||dj ƒ}t   |¡|}t|ƒd}|D]:} | j dkršqˆ|| j } | ddkr²qˆt| j d¡ƒ} |D]ü} | | krÔqÆt  t  | j| j¡¡} | |krôqÆt| t| j d¡ƒƒ}| jd|…}| jd|…}||k}t  ||d k||d k¡}| ¡}|dkr‚t  t  | jd|…| jd|…¡¡} | |}| |}||kr¢d}n|dkr´||}| j |8_ qÆqˆd d „|Dƒ}| d t|ƒ¡|S) aDeblur the reads Parameters ---------- input_seqs : iterable of (str, str) The list of input sequences in (label, sequence) format. The label should include the sequence count in the 'size=X' format. mean_error : float, optional The mean illumina error, used for original sequence estimate. Default: 0.005 error_dist : list of float, optional A list of error probabilities. The length of the list determines the amount of hamming distances taken into account. Default: None, use the default error profile (from get_default_error_profile() ) indel_prob : float, optional Indel probability (same for N indels). Default: 0.01 indel_max : int, optional The maximal number of indels expected by errors. Default: 3 Results ------- list of Sequence The deblurred sequences Notes ----- mean_error is used only for normalizing the peak height before deblurring. The array 'error_dist' represents the error distribution, where Xi = max frequency of error hamming. The length of this array - 1 limits the hamming distance taken into account, i.e. if the length if `error_dist` is 10, sequences up to 10 - 1 = 9 hamming distance will be taken into account NzUsing error profile %szno sequences deblurredzdeblurring %d sequencesrrgš™™™™™¹?ú-écSsg|]}t|jƒdkr|‘qS)r)Úroundrrrrr r½szdeblur..z-%d unique sequences left following deblurring)rrrr Údebugr(rÚinforÚpowrÚnpZarrayrZsequenceÚrstripZ count_nonzeroZ not_equalZ np_sequenceÚminZ logical_orÚsum)r%Z mean_errorr Z indel_probZ indel_maxr'r&Z mod_factorZ max_h_distZseq_iZnum_errZ seq_i_lenZseq_jZh_distrZ sub_seq_iZ sub_seq_jÚmaskZ mut_is_indelZ num_indelsZnum_substitutionsZcorrection_valueÚresultrrr ÚdeblurGsf$        ÿ ÿ  ÿÿ  r6)rNrr)) ÚoperatorrZnumpyr0rZdeblur.sequencerr r(r6rrrr Ú s   ,þ