DUffvddlZddlZddlZddlZddlZddlZddlZddlZddlZddl Z ddl Z ddl Z ddl Z n #e $rYnwxYwddlmZddlmZddlmZddlmZddlmZdZiZd5d Zd Zd5d Zd6dZdZGddeZGddeZGddeZdZ dZ!dZ"dZ#dZ$dZ%dZ&dZ'd7dZ(dZ) d8d!Z*d"Z+d#Z,d$Z-d%Z.ddlZej/d&ej0Z1d'Z2Gd(d)e3Z4Gd*d+e3Z5d9d,Z6 d:d0Z7d;d1Z8 d;d2Z9d3Z:d4Z;ej<e(dS)<N)settings) filenames)genome_registry)logger)create_interval_from_listc<ddlm}||dS)aY Explicitly set path to `BEDTools` installation dir. If BEDTools is not available on your system path, specify the path to the dir containing the BEDTools executables (intersectBed, subtractBed, etc) with this function. To reset and use the default system path, call this function with no arguments or use path="". rpathsN)r r _set_bedtools_pathpathr s O/var/www/html/software/conda/lib/python3.11/site-packages/pybedtools/helpers.pyset_bedtools_pathrs0 T"""""c6ddlm}|S)z4 Returns the currently-set path to bedtools rr )r r _get_bedtools_pathr s rget_bedtools_pathr.s(  # # % %%rc<ddlm}||dS)z Explicitly set path to `R` installation dir. If R is not available on the path, then it can be explicitly specified here. Use path="" to reset to default system path. rr N)r r _set_R_pathrs r set_R_pathr7s0 drFctjr|sdSttjdks|r t jt jtj ddg d }|rtd|zdt_||}tjd|}|s"td ||d }d |d Dt_tjdd kotjd dkt_tjdd kotjd dkt_dS#tj$r1tj rdtj z}nd}t-d|zwxYwdS)z Checks installation as well as version (based on whether or not "bedtools intersect" works, or just "intersectBed") Trbedtools --versionzutf-8zFound bedtools version '%s'Nz^bedtools [^0-9]*([0-9][0-9.]*)z(Cannot identify version number from "{}"rc,g|]}t|S)int.0is r z'_check_for_bedtools..hs(G(G(GAQ(G(G(Gr.(tried path '%s')r zmPlease make sure you have installed BEDTools(https://github.com/arq5x/bedtools) and that it's on the path. %s)r_bedtools_installedlenbedtools_version subprocess check_outputosrjoin_bedtools_pathdecoderstripprintresearch ValueErrorformatgroupsplit _v_2_27_plus _v_2_15_plusCalledProcessErrorOSError) force_checkverboseoverridevmvvadd_msgs r_check_for_bedtoolsrEEs #Kt H %&&!++ +- 'W\\("9:FF T  93a7888+/H (#  ;Q??A W !K!R!RST!U!UVVVB(G(G#(G(G(GH %)!,1Xh6OPQ6RVX6X  ! )!,1Xh6OPQ6RVX6X  ! ! !,   & -0GG')01  K ,+s E*FAGcF tjtjt jddgtjtj}dt _dS#t$r1t jrdt jz}nd}td|zwxYw)NRr)stdoutstderrTr(r z1Please install R and ensure it is on your path %s) r,Popenr.rr/r_R_pathPIPE _R_installedr=r6)prDs r _check_for_RrO~s X   W\\(*C 0 0+ >??    !% XXX   )H,<>' " "(@7J'''Hrc(tjS)z2 Gets the current tempdir for the module. )r gettempdirrrr get_tempdirrs    rctjrdStjD]G}|rt d|t j|rt j|H|rWtj t j td}|D]}t j|dSdS)z Deletes all temp files from the current session (or optionally *all* sessions) If *verbose*, reports what it's doing If *remove_all*, then ALL files matching "pybedtools.*.tmp" in the temp dir will be deleted. Nremovingzpybedtools.*.tmp) rKEEP_TEMPFILESr TEMPFILESr3r.rrunlinkglobr/r)r? remove_allrifnss rcleanuprs!  " *b ! ! ! 7>>"    IbMMMi []]4FGGHH  B IbMMMM  rc4tjsttjs|gS tj|}n0#t $r#|tjvr t||dzwxYwtj tj d|gS)Nz!not a recognized BEDTools programr) rr)rEr; _prog_namesKeyError _new_namesrYr.rr/r0) prog_names r_version_2_15_plus_namesr-s  '  {X(3 XXX + + + Iy3V'VWWWX GLL0* = =y IIs >-A+Tc` |du}|du}|dd}g}|t|d||dd |r|r tjdtjdd|t j|t jt jt jt} |r2|D].} | j | /n|D]} | j | | j |rd| j D} nd | j D} d} |r|stjd tjdd|t|d } t j|| t jt jt} t|d r+| |\}} nH|D].}| j | /| \}} |} | |s|rtjd tjdd|t j|t jt jt} |rd| j D} nd| j D} d} |s|stjd|t%t't(|}tjdd|t|d } t j|| t jt} | \}} |} | |Vt+| t,r| d} || r!t0j | d} | rt+| t,r| d} t5| dkr>d| ddvr t0j | n"t9t j|| n#t<t>f$r}tAtC|dtEj#|j$tAdt j|zdd|zfddd}tAdtAdd||j$zt=dd}~wwxYw| S)a Use subprocess.Popen to call BEDTools and catch any errors. Output goes to *tmpfn*, or, if None, output stays in subprocess.PIPE and can be iterated over. *stdin* is an optional file-like object that will be sent to subprocess.Popen. Prints some useful help upon getting common errors. *check_stderr* is a function that takes the stderr string as input and returns True if it's OK (that is, it's not really an error). This is needed, e.g., for calling fastaFromBed which will report that it has to make a .fai for a fasta file. *decode_output* should be set to False when you are iterating over a BAM file, where the data represent binary rather than text data. Nrrz:helpers.call_bedtools(): input is stream, output is streamz helpers.call_bedtools(): cmds=%s )rHrIstdinbufsizec3@K|]}|dVdSUTF-8Nr1r s r z call_bedtools..w.>>!((7++>>>>>>rc3K|]}|VdSr[rr s rrz call_bedtools..y"..!......rz8helpers.call_bedtools(): input is stream, output is filewbrhz.rrc3K|]}|VdSr[rr s rrz call_bedtools..rrzChelpers.call_bedtools(): input is filename, output is filename (%s)UTF_8WARNINGz: zThe command was: %s )z&* Did you spell the command correctly?z1* Do you have BEDTools installed and on the path?z<* Do you have permission to write to the output file ("%s")?)zM* Too many files open -- please submit a bug report so that this can be fixed)z* Broken pipe -- if you passed a BedTool object that was created using a generator function, please try saving it to disk first using the .saveas() method before calling this bedtools command. See issue #49 for more.)r%  zThings to check:z z*See above for commands that gave the error)%rrrdebugr/r,rJrLBUFSIZErwriteencodecloserHrghasattr communicaterhrmapr\rbytesr1sysrIr*upperrY list2cmdliner=IOErrorr3typer.strerrorerrno)cmdstmpfnr check_stderr decode_output encode_inputinput_is_streamoutput_is_stream _orig_cmdsrNlineoutputrIoutfilerHrerrproblemss r call_bedtoolsr;s64'O}aaaJ DKK(A77888KK 122HD  /  LO    L;SXXd^^ L L L !! o A (!11DGMM$++--00001"((DGMM$'''' GMMOOO />>QX>>>..QX...F  #3  LS T T T L;SXXd^^ L L L5$''G ! o Auf%% 1!"uzz||!>QX>>>..QX...F '7  L#    C''D L;SXXd^^ L L L5$''G WZ_gA]]__NFFF MMOOO  #&%(( 0w//|F##    (((  K&%(( 0w//6{{RI1B1B1D1D$D$D   ((((#J$;D$A$A6JJJ W DDD $s))))R[%;%;%;<=== *Z-DT-J-JJKKK -/45  ,  !!! fv{{8CI#6777888BCCC9D< MsRS V+1B5V&&V+ct|tr|d}|drdS|drdSdS)zj If stderr created by fastaFromBed starts with 'index file', then don't consider it an error. rz index fileTrF)rrr1 startswith)rs r_check_sequence_stderrrs^ !U HHW  ||L!!t||It 5rc Vt||||||dd|S)z Helper function that list-ifies the output from randomintersection, s.t. it can be pickled across a multiprocess Pool. FN)intersect_kwargsshuffle_kwargsreport_iterationsr processes_orig_processes)rrandomintersection)_selfother iterationsrrrrrs r_call_randomintersectrsF     -)/+ !    rc<|D]}t|jtrtj|jn.t |jdr|jt |jdr|jtdS)zw Single function that can be used to get rid of a BedTool, whether it's a streaming or file-based version. rthrowN) rrir\r.rrrr StopIteration)argsrs rclose_or_deleters && adC   IadOOOO QT7 # #  DJJLLL 14 ! ! & DJJ} % % % &&rc tj}tjddddt |g}d}|D]/}|ddr|ddkr|dz }0|S)Nlsofz-wz-Ffz-prrrj)r.getpidr,r-r\ splitlinesisdigit)pidprocsnprocsr"s r n_open_fdsr #s )++C  #VT5$C$I J JE F      QRR5==?? qts{{ aKF MrzT (?P.+): (?P\d+)- (?P\d+) (?:\[(?P.)\])?c t|trt|}|drat |d|d|ddd|dgSt |d|d|dgS|S)z Convert string of the form "chrom:start-stop" or "chrom:start-stop[strand]" to an interval. Assumes zero-based coords. If it's already an interval, then return it as-is. strandchromrkstopr$0)rr\coord_rer5r8r)srBs rstring_to_intervalr9s!S OOA   778   ,GGG$$GGG$$GGFOOGGH%%    -!!1777#3#3QWWV__E  Hrc eZdZdZdZdZdS) FisherOutputc t|tr!t|}t |drdd|D}ddddddd}||_|}|D]}d|vr\| d\}}}}t||dd <t||dd <d |vr\| d\}}}}t||d d <t||d d <||_ |d  \} } } } t| |_ t| |_t| |_t| |_dS)a fisher returns text results like:: # Contingency Table #_________________________________________ # | not in -b | in -b | # not in -a | 3137160615 | 503 | # in -a | 100 | 46 | #_________________________________________ # p-values for fisher's exact test left right two-tail ratio 1.00000 0.00000 0.00000 2868973.922 nextr c3K|]}|VdSr[rr s rrz(FisherOutput.__init__..is~~a~~~~~~rN) not in -bin -b) not in -ain -ar|rrz in -arr )rr\rgrhrr/textrstripr9rtablefloat left_tail right_tailtwo_tailratio) r_rkwargsr!linesr"_in_bnot_in_bleftrightr%r&s rr`zFisherOutput.__init__Ws a   Q A 1f   &~~1~~~%%A'+d;;#'$77     4 4Aa'(wwyys';';$4125h--k";/.1$iik"7+a'(wwyys';';$41.1(mmg{+*-d))gw' ',Ry'8'8$eXut,,h 5\\ rc|jSr[)rr_s rrczFisherOutput.__str__s yrcPd|jjdt|d|jS)N ) __class__rRidrr/s r__repr__zFisherOutput.__repr__s*#'>#:#:#:BtHHHHdiiPPrN)rRrSrTr`rcr4rrrrrVsG("("("TQQQQQrrceZdZdZdS) SplitOutputc ddlm}t|tr!t |}t |drdd|D}||_g|_ g|_ g|_ g|_ | D]}|}|j |d|j t!|d|j t!|d|j ||ddS) a) Handles output from bedtools split, which sends a report of files to stdout. This class parses that list into something more convenient to use within pybedtools. Most useful is probably the .bedtools attribute, which is a list of BedTool objects. r)BedToolrr c3K|]}|VdSr[rr s rrz'SplitOutput.__init__..s"//1Q//////rrr%N)bedtoolr8rr\rgrhrr/rrfilesnbasescountsrr9rr)r_rr'r8rtokss rr`zSplitOutput.__init__sQ %$$$$$ fc " " )&\\&&((F 66 " " 0WW///////F     %%'' 3 3D::<))*BG)TTt <      5s!%AAmysqlfetchChromSizesgenome-mysql.cse.ucsc.educt|std|dd|zddd|zg}g}i} tj|tjtjt } | \} } | r?t| td ttj|| d d } | D]W} t| tr| d } | \}}d t|f||<X|t||nL#t $r?}|jdkr)|d|nYd }~nd }~wwxYw ||g}tj|tjtjt } | \} } | rPdt)| vr?t| td ttj|| } | D]W} t| tr| d } | \}}d t|f||<X|t||n7#t $r*}|jdkr|dYd }~nd }~wwxYw|st!||S)a Download chrom size info for *genome* from UCSC and returns the dictionary. Parameters ---------- genome : str Name of the genome assembly (e.g., "hg38") saveas : str Filename to save output to. Dictionary will still be returned. mysql, fetchchromsizes : str Paths to MySQL and fetchChromSizes. timeout : float How long to wait for a response; mostly used for testing. host_url : str URL of UCSC mirror MySQL server. r@zVIt appears you don't have an internet connection -- unable to get chromsizes from UCSCz --user=genomez--host=z-Az-ez$select chrom, size from %s.chromInforzCommands were: rNrrr%zCan't find mysql at path {0}zINFO: trying WGETz"Can't find path to fetchChromsizes)rHr6r,rJrLrrr3rrrrr1r9rchromsizes_to_filer=rrr7r\)genomesaveasrIfetchchromsizesrAhost_urlrfailuresdrNrHrIr(rrsizers rget_chromsizes_from_ucscrUs8: w ' ' '  4   H  .7  DH A   '     3 &MMM $ % % % :*400 2 2 2!!##ABB' & &D$&& ,{{7++**,,KE43t99~AeHH   q& ) ) )  9>> OO:AA%HH I I I I  J I I I I B(   '     7"#f++55f ()))z.t44666!!## & &D$&& ,{{7++**,,KE43t99~AeHH   q& ) ) ) BBB 9>> OO@ A A AB  h Hs1DE F5F  FDJ11 K%; K  K%c|?tjddd}|j}tj||}t |trt|}t|d}| D]8\}}|dzt|dzd z}| |9| |S) z Converts a *chromsizes* dictionary to a file. If *fn* is None, then a tempfile is created (which can be deleted with pybedtools.cleanup()). Returns the filename. Nz pybedtools.z.tmpF)prefixsuffixdeletewt rrb) rNamedTemporaryFilenamerrrrr\ chromsizesrgrrr) chrom_sizesrirfoutrboundsrs rrMrMs z+     ""5))) +s##. -- D>>D$**,, vt|c&)nn,t3 4JJLLL IrcDdtjvrdSi} tj|}t |dsdSt |jD]-}|\}}dt|f||<.|t||n#t$rYdSwxYw|S)a Get chrom size info for *genome* from genomepy, if genomepy is installed. Parameters ---------- genome : str Name of the genome assembly (e.g., "hg38") saveas : str Filename to save output to. Dictionary will still be returned. genomepyN sizes_filer) rmodulesrcGenomerrgrdr9rrMr)rNrOrSgrrrTs rget_chromsizes_from_genomepyrh5s$$t A  OF # #q,'' 4&& & &D**,,KE43t99~AeHH   q& ) ) ) tt Hs$BAB BBc tt|S#t$r't|}|t |cYS|cYSwxYw)a| Looks for a *genome* already included in the genome registry; if not found it first tries to look it up via genomepy. If genomepy is not installed, or if this lookup fails then it looks it up on UCSC. Returns the dictionary of chromsize tuples where each tuple has (start,stop). Chromsizes are described as (start, stop) tuples to allow randomization within specified regions; e. g., you can make a chromsizes dictionary that represents the extent of a tiling array. Example usage: >>> dm3_chromsizes = chromsizes('dm3') >>> for i in sorted(dm3_chromsizes.items()): ... print(i) ('chr2L', (0, 23011544)) ('chr2LHet', (0, 368872)) ('chr2R', (0, 21146708)) ('chr2RHet', (0, 3288761)) ('chr3L', (0, 24543557)) ('chr3LHet', (0, 2555491)) ('chr3R', (0, 27905053)) ('chr3RHet', (0, 2517507)) ('chr4', (0, 1351857)) ('chrM', (0, 19517)) ('chrU', (0, 10049037)) ('chrUextra', (0, 29004656)) ('chrX', (0, 22422827)) ('chrXHet', (0, 204112)) ('chrYHet', (0, 347038)) )getattrrrrhrU)rNr^s rr^r^YsfD/// 1&99  +F33 3 3 3    s*AAActjtjtjt }|tj|dgS)zE Returns a list of include directories with BEDTools headers include)r.rabspathr/dirname__file__)rns r get_includesrpsNgoobgll27??8+D+DEEFFG RW\\'955 66r)r )FFN)FF)NNNTT)r)NrIrJNrKr[)=rr.rrr,rruatexitr4rB urllib.errorurllib.requestrc ImportErrorr rrrr cbedtoolsrrrrrrrErO ExceptionrQrWrYrlrrrrrrrrrrrrrr compileVERBOSErrobjectrr6rHrUrMrhr^rpregisterrrrr{s,   OOOO   D 000000   # # # # &&&    6666r X X X      I        e        E    4     1 1 1   "   !!!. J J J   mmmm`   8 & & & 2: J      :/Q/Q/Q/Q/Q6/Q/Q/Qd%3%3%3%3%3&%3%3%3P  %  ( a a a a H2! ! ! ! H)))X777s 7??