tf7<`ddlZddlZddlZddlmZddlZddlZddlZ ddl m Z ddl Z ddl Z ddl mZddlmZddlmZddlmZddlmZdBdZdZej:dZdZejAe jBd d d e jDd ddde jDddde jFejHe jDddde%e jDddde%e jDddde%e jDddd e%e jDd!d"d#$e jDd%d&d'$e jDd(d)e&d*e jDd+d,d$e jDd-d.d$e jDd/d0d$e jDd1e jFgd2d3d45e jDd6d7d$e jDd8d9d:d;e jDdd?gd>*d@Z'ejAe jBd d d e jDd ddde jDddde jFejHe jDd+d,d$e jDd6d7d$e jDd8d9d:d;dAZ(y)CN)tqdm)clic6|d}|jjd}|djdd}ttj|d}|Dcgc]}|d }}|d}t |d} t |d } || | |fScc}w) aM Convert a line from an epilogos bedfile to vector format. Parameters ----------- bedline: [string,....] A line from a bedfile broken up into its constituent parts (e.g. ["chr1", "1000", "2000", "[1,2,34,5]"]) Returns ------- An array containing the values associated with that line r :c |dS)Nrxs ]/var/www/html/software/conda/envs/higlass/lib/python3.12/site-packages/clodius/cli/convert.pyz,epilogos_bedline_to_vector..)s !A$)keyr)stripsplitsortedast literal_evalint) bedlines row_infosbedlineparts array_str array_valvstateschromstartends repilogos_bedline_to_vectorr%sqkG MMO ! !$ 'Eas#B'Is'' 2GI% &qad &F & !HE aME eAh-C 5#v && 's! Bc|d}|jjd}|d}t|d}t|d}||d}tt |Dcgc] }||k(rdnd } }|||| fScc}w)a Convert a line from a bedfile containing states in categorical data to vector format. Parameters ---------- bedline: [string,...] A line form a bedfile broken up into its contituent parts (e.g. ["chr1", "1000", "2000", "state"])) states_dic: {'key':val,...} A dictionary containing the states in the file with a corresponding value (e.g. {'state1_name': 1, 'state2_name': 2,...}) Returns ------- Four variables containing the values associated with that line: chrom, start, end, states_vector (e.g. chrom = "chr1", start = 1000, end = 2000, states_vector = [1,0,0,0]) rrrrr)rrrrangelen) r states_dicrrr"r#r$stateindex states_vectors rstates_bedline_to_vectorr-3s2qkG MMO ! !$ 'E !HE aME eAh-C uQx E=B3z?=STE%5.Qa/TMT 5#} --Us$A<cy)zW Aggregate a data file so that it stores the data at multiple resolutions. Nr r rrconvertr/Xs  rc D tdtj5}td|tj|d}t j |d}tj| |\}}}|Bt|d5}|Dcgc]!}|jjd#}}dddnd}|jD]V}|j|tj|j || z  t#|zft$j&dXd fd }| d k(r t)j*||t,| || n| d k(rJd |Dcgc]%}|j/dj1d d'}}t3t#|Dcic]}||| }}t)j*||t4| || |nt)j*|||| || |j7|} t j | d}!|tj8|dddz}td|tj:|rt=j>||dk(rd}"nd}"| d k(rED#cgc]}#|#jd}$}#t)j@|!tC|||"| |||$n&t)j@|!tC|||"| ||dddycc}w#1swYExYwcc}wcc}wcc}#w#1swYyxYw)Nz chrom_col:temporary dir:temp.mv5wrutf8gzip fillvalue compressionct}t}t}g}|D] }|jj}| dz }t|dz } t|dz } |dz dz zD cgc]"} | dk(s t | nt j $} } |j||j| |j| t|dkDr td|t|dkDr td|t|dkDr td||| z } t|dt|dt|d|fScc} w)NrNAz'Chromosomes don't match in these lines:z+Start positions don't match in these lines:z)End positions don't match in these lines:r) setrrrfloatnpnanaddr( ValueErrorlist)rr chrom_set start_setend_set all_vectorrrr"r#r$fvector chrom_col from_pos_colnum_rows to_pos_col value_cols r!bedline_to_chrom_start_end_vectorz@_bedgraph_to_multivec..bedline_to_chrom_start_end_vectorsIIeGJ# % --/i!m,E,"234% Q/0#9q=9q=83KL%&IE!H2669 e$ e$ C y>A%$A8y>A%$ExwnanminrAisnansmrPr?)ra orig_shapena SMALL_NUMNAN_THRESHOLD_NUMresnress raggz"_bedgraph_to_multivec..aggsCCKKR 34WW YYu% $-O!99R=#44$Q$-288B< ZZ +ll11-//{{5)13T--.ll399- rc|jj|jdddfjdjSNrr rrSrUrVrWsumr s rrbz"_bedgraph_to_multivec..aggs8ss{{AGGAJA#67;;;CEEEr chromsizesrbstarting_resolution tile_size output_filerN)"printtempfileTemporaryDirectoryopjoinh5pyFilecchload_chromsizesopenrencode chrom_ordercreate_datasetmathceil chrom_lengthsr(r>r?cmvbedfile_to_multivecr%decoderr'r-closesplitextexistsosremovecreate_multivec_multireszip)% filepathsrkassemblyrIrJrLrM has_header chunk_size nan_valuechromsizes_filenamerirKformatrow_infos_filenamerjmethodtd temp_filef_out chrom_info chrom_names chrom_sizesrGlinerr"rNlne states_namesrr)tff_inrb state_namestates_row_infoss% ```` ` r_bedgraph_to_multivecrasV& , "  $ $ &s" #GGB +  )S)141D1D 2 .[+  )(#. H!EFGTTZZ\008G G H HI++ E  IIj66u=@SSTs9~-&&" !  & & P Z   # #*# x % LK L%IRS#CJJv.44T:1=SLS6;C N6KL,q/1,LJL  # #(#   # #1#   yyS!  ++il3A6HK nk* 99[ ! IIk " [  D F X The column number (1-based) which contains the chromosome name)rrrz--from-pos-colz@The column number (1-based) which contains the starting positionrz --to-pos-colz>The column number (1-based) which contains the ending positionrz --value-colz;The column number (1-based) which contains the actual valuez--has-header/--no-headerz2Does this file have a header that we should ignoreF)rrz --chunk-sizez)The size of the chunks to read in at oncegj@z --nan-valuez The string to use as a NaN value)rrrz--chromsizes-filenamez,A file containing chromosome sizes and orderz--starting-resolutionzbThe base resolution of the data. Used to determine how much space to allocate in the multivec filez --num-rowsz:The number of rows at each position in the multivec formatz--format)rrOr!z'default':chr start end state1_value state2_value, etc; 'epilogos': chr start end [[state1_value, state1_num],[state2_value, state2_num],[etc]]; 'states': chr start end state_namer)rrrz--row-infos-filenamez.aggs833;; B2377Q7?AA Arrg)#rnrormrprqrrrsrtrurvrrwrxryrzr{r|r(r>r?rrB enumeratebbi is_bigwigrhr<keys intersectionfetchflushrr}rr)rrkrrrrjrrrrrr|rGrrri resolutionr"bw_indexbw_filerhmatching_chromosomeschr_namechr_len chr_shapearrrrrbs rbigwigs_to_multivecrsT  $ $ &C " #GGB +  )S)363F3F 4 0[-  )(#. H!EFGTTZZ\008G G H HI( ++ E  IIj66u=@SST N&&" !  "&d9Y+?&@y!Q 2 Hg}}W% ^^G4 '*:??+<'='J'J $($ !57H+x<(66x@G!%7Z+?!@#i. QI))1gy|UC47E(OAxK07 01# 2&   yyS! B $$ ; 6 3# wC C H H HC C s7A!J6J;&I?!J#GJ?JJ JJrl))rrzros.pathpathrprnrrnumpyr>rrclickclodius.chromosomes chromosomesrtclodius.multivecmultivecr}negspy.coordinates coordinatesnc scipy.miscmiscrZrr%r-groupr/rcommandargumentoptionChoiceavailable_chromsizesrstrr~rr rrrs   !'<".J    HV  [;     A -b--/ 0   I   K   I   F   =  Dc:d 7      E   7 8 )   G      E uk* +   %           <|%P  [;     A -b--/ 0   7   G     K    <DK r