DUf<@ddlZddlZddlZddlZddlmZmZmZm Z m Z m Z ddl m Z dZddZdZdZddZdd Zd Zd ZGd d eZedkr edSdS)N)Fasta wrap_sequence FetchError ucsc_split bed_splitget_valid_filename) defaultdictc tjj\}}|r |dd}t jjj}jrfd}tjj tj j j|jj }t#\}}|s|}d}|D]}||\} } } | dkrjrB| | | | z } nt)|| } jd| ksjd| krbjrCdd| | | |fD} t/| } t1| d}njrj}n t4j} jrK|s"jd kr|d d }|t=|| | | n+t?|| | | D]}||n1#t@$r$}tAtC|d zd}~wwxYwjr|"|#dS) Nc^tjj| SN)recompileregexsearch)xargss H/var/www/html/software/conda/lib/python3.11/site-packages/pyfaidx/cli.pyz write_sequence..s$bj&<&<&C&CA&F&F"F) default_seq key_function strict_bounds split_char filt_functionread_long_namesrebuildFr.c38K|]}|t|VdSr )str).0es r z!write_sequence..'s-MM11MAMMMMMMrw nucleotidez name start end A T C G N others Tz Try setting --lazy. )$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 W  dj ) )FAs !""gJtz**1M GFFFF $*$*:dNbIcIcw{xAtANRN\lyKOKZhlhwdw x x xE'4T':':$n ( ::<< F"  )>&11eS 4<<  ?  S_"U{ "5;// q!L00DOA4F4U4U   !xxMMuc30GMMMMMH)(33H8S))GG X !hGGjG @~ ("$.L"@"@MM"MNNN!F 0udE3OOPPPP*4eSII((DMM$'''' @ @ @SVV&>>?? ? @    MMOOO NNsA=H I$IIc #xK |jj|j}|jr-||kr'|%|#|||dz |dz}|jj}n||||}nG#t $r:tj dj ditYdSwxYw|jr|j}|jr|j}|j rdS|j rnJ|s|r'|js dd|jdgVndd|jdgVt%||jD]}|VdS)Nr z"warning: {name} not found in file > )faidxindexlenc auto_strandreverse complementKeyErrorr9stderrr<formatlocals no_outputno_names no_coordsr6 fancy_namerGrseq)rr)rGrHrIline_lensequencerMs rr>r>>s ;$T*/   . 0AcoT{37519#45H'2HHT{59-H  E>EQQQQRRR '& |$# ~ }6  6S 6$. 6''3 3T:;; ; ; ; ;''3 t455 5 5 5h 55 sAA AB$#B$ct|jd|j}t|\}}|D]}||\}}}|jrW|r|r||z }n7|s|st ||}nt ||||}||jz||||<o|jr-||||||||<dS)NT)mutabler) rr)r.r1mask_with_default_seqr4r mask_by_case lowercase) rr)rCrDrFrnamerHrIspans r mask_sequencerl[s $*dt~ F F FE'4T':':$n" J J*N622uc  % J 4 4U{ 43 45<((5<c 233&*T-=&=E%Ls # #   J&+El59&=&G&G&I&IE%Ls # J JrcR|jr|j}t}n|j}t}||fSr )bedrregionsr)rrCrDs rr1r1ns3 x$8"<# n --rcf|jj|j}||||}|jr|j}|jr|j}|jrdS|jdkr*d|j|j dz |j S|jdkr)d|jt|S|jdkr.t| tt}|fd t#D|d d }|d d } |d d } |dd } |dd } dd|D} djd|j|j |j dt+S|jdkr5d|j|j |j t|SdS)Nrnz{name} {start} {end} r )rGrHrI chromsizesz{name} {length} )rGlengthr%c>g|]}||fSrS)count)r!csss r z&transform_sequence..s(777!a!%777rArTCGN|c\g|])\}}d|t|f*S):)r6r )r!kvs rrwz&transform_sequence..s2JJJTQ388QAK00JJJrz5{sname} {sstart} {send} {A} {T} {C} {G} {N} {others} )snamesstartsend transposedz{name} {start} {end} {seq} )rGrHrIrbrS)rTrUrVrYrXr^r;r\rGrHrIr4r upperr intupdatesetpopr6itemsr])rr)rGrHrIrcsnucsrxryrzr{r|othersrvs @rr=r=xs3{ &+H d E#IA  L | I ~ ~)00afAGaKVWV[0\\\ < ' '#**s1vv*FFF < ' ' VV\\^^3 7777s2ww777888 HHS!   HHS!   HHS!   HHS!   HHS!  JJTZZ\\JJJKKVOVM]^]clmlsz{zMMDJDLDLMM M < ' '077QV17XYX]cfghcici7jjj ( 'rcddlm}tjdd}|dt d|d t d d |d }|d}|d}|ddtjdd|ddtjdd|ddt dd|dddd d!"|d#d$dd d%"|d&d'dd d("|d)d*tdd+,| }|d-d.dd d/"|d0d1dd d2"|d3d4dd d5"|d6d7dd d8"|d9d:dd d;"|d,|d?d@t ddA,|dBdCt dDdE,|dFdGt dHdIdJK|dL}|dMdNt dOdP,|dQdRdd dS"| }|dTdUdd dV"|dWdXdd dY"|dZdd d["|d\dd d]"|d^d_|d`attj dbkr+|s)|tjdbn,|r||} n|} | jrL| jrtjdc| jrtjdd| js| jrt1| dSt3| dS)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%rzItransform 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#lK|D]\}}td|D]}|VdS)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)relemrtrAs relementszCounter.elementssX >>++  KD%D%((      rc td)Nz@Counter.fromkeys() is undefined. Use Counter(iterable) instead.)NotImplementedError)clsrrs rfromkeyszCounter.fromkeyss! NPP Prc <|t|drR|r4|j}|D]\}}||d|z||<n:t||n|j}|D]}||ddz||<|r||dSdS)zLike dict.update() but add counts instead of replacing them. Source can be an iterable, a dictionary, or another Counter instance. Nrrr )hasattrgetrdictr)rrrself_getrrts rrzCounter.update s  x-- 70#xH'/'9'9';';?? e%-XdA%6%6%>T ?KKh////8$77D!)$!2!2Q!6DJJ   KK       rc t|S)zBLike dict.copy() but returns a Counter instance instead of a dict.)r)rs rcopyz Counter.copy st}}rcH||vrt||dSdS)zGLike dict.__delitem__() but does not raise KeyError for missing values.N)r __delitem__)rrs rrzCounter.__delitem__$s. 4<<   T4 ( ( ( ( ( rlr1r=rrrrrrrSrrrsR  ________________######222j:JJJ&...kkkk:5555r"""MMMMMdMMM^ zDFFFFFr