tf<@ddlZddlZddlZddlZddlmZmZmZm Z m Z m Z ddl m Z dZddZdZdZddZdd Zd Zd ZGd d eZedk(reyy)N)Fasta wrap_sequence FetchError ucsc_split bed_splitget_valid_filename) defaultdictc tjjj\}}|r|dd}t j j j}jrfd}tjjtjj j|jj }t#\}}|s|j%}d}|D]Z}||\} } } | dk(rj&r=| | | | z } nt)|| } j&d| kDsj&d| kr_j*r4dj-d| | | |fD} t/| } t1| d}n)j2r j2}nt4j6} j8rC|s"j8d k(r|j;d d }|j;t=|| | | n%t?|| | | D]}|j;|j*sK|jE]|jGy#t@$r}tAtC|d zd}~wwxYw) Ncbtjjj| SN)recompileregexsearch)xargss U/var/www/html/software/conda/envs/higlass/lib/python3.12/site-packages/pyfaidx/cli.pyz write_sequence..s"bjj&<&C&CA&F"F) default_seq key_function strict_bounds split_char filt_functionread_long_namesrebuildFr.c38K|]}|st|ywr )str).0es r z!write_sequence..'sM11AMsw nucleotidez name start end A T C G N others Tz Try setting --lazy. )$ospathsplitextfastarrrr invert_matchrrevalheader_functionlazy delimiter long_names no_rebuild split_regionskeys size_rangelen split_filesjoinropenoutsysstdout transformwritetransform_sequencefetch_sequencerr close__exit__)r_extrr)regions_to_fetchsplit_functionheaderregionnamestartend sequence_lenfilenameoutfileliner"s` rwrite_sequencerN s WW  djj )FAs !"gJJtzz*11M F $**$*:*:dNbNbIcw{xAxAtANRN\N\lyKOKZKZhlhwhwdw xE'4T':$n  ::< F" )&1eS 4<  ?? S_"U{ "5;/ q!L0DOOA4F4U   xxMuc30GMMH)(3H8S)G XXhhGjjG @~~$..L"@MM"MN!F 0udE3OP*4eSI(DMM$'(    MMOA B NN  @SV&>>? ? @sA4I&& J /JJ c #K |jj|j}|jr.||kDr)|'|%|||dz |dz}|jj }n||||}|j r |j }|jr |j}|jry|jrnQ|s|r-|js!djd|j dgn djd|j"dgt%||j&D]}|y#t $r:tjjdjditYywxYww)Nr z"warning: {name} not found in file > )faidxindexlenc auto_strandreverse complementKeyErrorr9stderrr<formatlocals no_outputno_names no_coordsr6 fancy_namerGrseq)rr)rGrHrIline_lensequencerMs rr>r>>sI ;;$$T*//    0AcoT{375195H''22HT{5-H && ||## ~~ }} S$..''3 3 3T:; ;''3 t45 5h 5 !  E>EEQQRs+E*A%D$)B;E*$AE'$E*&E''E*c|t|jd|j}t|\}}|D]}||\}}}|jrD|r|r||z }n$|s|st ||}nt ||||}||j z||||_|jsl||||j||||y)NT)mutabler) rr)r.r1mask_with_default_seqr4r mask_by_case lowercase) rr)rCrDrFrnamerHrIspans r mask_sequencerl[s $**dt~~ FE'4T':$n" J*62uc  % %U{35<(5<c23&*T-=-=&=E%Ls #   &+El5&=&G&G&IE%Ls # Jrcr|jr|j}t}||fS|j}t}||fSr )bedrregionsr)rrCrDs rr1r1nsA xx88" n -- <<# n --rc |jj|j}||||}|jr |j}|jr |j}|j ry|j dk(r5dj|j|jdz |jS|j dk(r&dj|jt|S|j dk(rCt|j}tt}|j!t#|D cgc]} | |j%| fc} |j'd d } |j'd d } |j'd d } |j'd d } |j'dd }dj)|j+Dcgc]!\}}dj)|t|f#c}}}djd|j|j|jdt-S|j dk(r,|j d?d@t ddA,|j dBdCt dDdE,|j dFdGt dHdIdJK|j dL}|j dMdNt dOdP,|j dQdRdd dS"|j}|j dTdUdd dV"|j dWdXdd dY"|j dZdd d["|j d\dd d]"|j d^d_|d`attjdbk(r(|s&|jtjdbn$|r|j!|} n|j!}  j"rV| j$rtj&j)dc| j*rtj&j)dd| j,s | j.r t1| yt3| y)eNr) __version__zFetch sequences from FASTA. If no regions are specified, all entries in the input file are returned. Input FASTA file must be consistently line-wrapped, and line wrapping of output is based on input line lengths.zPlease cite: Shirley MD, Ma Z, Pedersen BS, Wheelan SJ. (2015) Efficient "pythonic" access to FASTA files using pyfaidx. PeerJ PrePrints 3:e1196 https://dx.doi.org/10.7287/peerj.preprints.970v1) descriptionepilogr)z FASTA file)typehelpro*z=space separated regions of sequence to fetch e.g. chr1:1-1000)rnargsrz input optionszoutput optionszheader optionsz-bz--bedrz1bed file of regions (zero-based start coordinate)z-oz--outr$z"output file name (default: stdout)z-iz --transform)rnrqr%r}zItransform the requested regions into another format. default: %(default)s)rchoicesrz-cz --complement store_trueFz-complement the sequence. default: %(default)s)actiondefaultrz-rz --reversez*reverse the sequence. default: %(default)sz-yz --auto-strandzQreverse complement the sequence when start > end coordinate. default: %(default)sz-az --size-rangezZselected sequences are in the size range [low, high]. example: 1,1000 default: %(default)s)rrrz-nz --no-namesz5omit sequence names from output. default: %(default)sz-fz --long-nameszpoutput full (long) names from the input fasta headers. default: headers are truncated after the first whitespacez-tz --no-coordszOomit coordinates (e.g. chr:start-end) from output headers. default: %(default)sz-xz --split-fileszEwrite each region to a separate file (names are derived from regions)z-lz--lazyz>fill in --default-seq for missing ranges. default: %(default)sz-sz --default-seqzDdefault base for missing positions and masking. default: %(default)sz-dz --delimiterzjdelimiter for splitting names to multiple values (duplicate names will be discarded). default: %(default)sz-ez--header-functionzlambda x: x.split()[0]z]python function to modify header lines e.g: "lambda x: x.split("|")[0]". default: %(default)sz-uz--duplicates-actionstop)rfirstlastlongestshortestzQentry to take when duplicate sequence names are encountered. default: %(default)s)rrrrzmatching argumentsz-gz--regexz.*zNselected sequences are those matching regular expression. default: %(default)sz-vz--invert-matchzRselected sequences are those not matching 'regions' argument. default: %(default)sz-mz--mask-with-default-seqz>+A??rc#fK|jD]\}}td|D]}|yw)zIterator over elements repeating each as many times as its count. If an element's count has been set to zero or is a negative number, elements() will ignore it. N)rrepeat)relemrrAs relementszCounter.elementss= >>+ KD%D%(    s/1ctd)Nz@Counter.fromkeys() is undefined. Use Counter(iterable) instead.)NotImplementedError)clsrrs rfromkeyszCounter.fromkeyss! NP Prc &|{t|drM|r4|j}|jD]\}}||d|z||<n9tj ||n"|j}|D]}||ddz||<|r|j |yy)zLike dict.update() but add counts instead of replacing them. Source can be an iterable, a dictionary, or another Counter instance. Nrrr )hasattrgetrdictr)rrrself_getrrs rrzCounter.update s  x-#xxH'/'9'9';? e%-dA%6%>T ?KKh/88$7D!)$!2Q!6DJ7  KK  rct|S)zBLike dict.copy() but returns a Counter instance instead of a dict.)r)rs rcopyz Counter.copy s t}rc:||vrtj||yy)zGLike dict.__delitem__() but does not raise KeyError for missing values.N)r __delitem__)rrs rrzCounter.__delitem__$s 4<   T4 ( rc|sd|jjzSdjtdj|j }|jjd|dS)Nz%s()z, z%r: %rz({z})) __class____name__r6map__mod__r)rrs r__repr__zCounter.__repr__)sSDNN333 3 #h..0@0@0BCD!^^44e<rlr1r=rrrrrrrSrrrsi  __#2j:J&.k:5r"MdM^ zFr