DUfddlZddlZddlZddlmZddlZejej  ddZ d Z d Z dd Z ddZdS)N)is_compatible_viewframe)level@Bstrandc#K|jd}tjtjd|||}t t |dd|dd}|D]\}}|||} | d|| d<| d|| d<| | ddk| ddkz} | jddkrQtj| ddg t| ddg<| dVtj d ||z d zd d dS) zXGenerates chunks of pixels from the cooler and adjusts their bin IDs to follow the view.rNrbin1_idbin2_idTdropz Processed dz.2fz% pixels)pixelsshapenpappendarangelistzipmapsortastypeint reset_indexlogginginfo) clr binmapping chunksizeorientation_colnprocnpixelschunksi0i1chunks T/var/www/html/software/conda/lib/python3.11/site-packages/cooltools/api/rearrange.py_generate_adjusted_chunksr) sjjll #G YryGY77 A AF #fSbSk6!"":.. / /F @ @B RU# +// ;;i +// ;;iuY'2-% 2Bb2HIJ ;q>A  ,.Gy),-44S99--E9i( )###.. . . . >"W*S.>>>>???? @ @ch|d|d<|d|dz |d<|S)Nlengthendstart)cumsum)chromdfs r(_adjust_start_endr1!s9X&--//GENu~(99GG Nr*c|jdddd}|d|dz }|d|dz |d<|d|z|d<|S)Nr Tr r-r.)ilocrmax)regdfls r( _flip_binsr7'sp Jttt  ( (d ( 3 3E e uW~%A5\%%''%,6E'N>A%E%L Lr* new_chromc tt|d||}tt|d||dk |dg}t j||ddg}|dd  fd d }t j ||d }|d ||d<|d|dz |d<|dd t ddg}tjdd|dt D}|dt|dt |jDtjd| dg}||fS)aZ Rearranges the input `bins_old` based on the information in the `view_df` DataFrame. Parameters ---------- bins_old : bintable The original bintable to rearrange. view_df : viewframe Viewframe with new order of genomic regions. Can have an additional column for the new chromosome name (`new_chrom_col`), and an additional column for the strand orientation (`orientation_col`, '-' will invert the region). new_chrom_col : str, optional Column name in the view_df specifying new chromosome name for each region, by default 'new_chrom'. If None, then the default chrom column names will be used. orientation_col : str, optional Column name in the view_df specifying strand orientation of each region, by default 'strand'. The values in this column can be "+" or "-". If None, then all will be assumed "+". Returns ------- bins_new : bintable The rearranged bintagle with the new chromosome names and orientations. bin_mapping : dict Mapping of original bin IDs to new bin IDs name-old_id)namesF)drop_unassigned view_region)subset) group_keysc@|jrt|n|S)N)r:r7)xflipdicts r(z rearrange_bins..Ws(16*:AAr*Tr )view_df df_view_colchromr-r.r,)columnszRearranged binsci|]}|dS)r ).0r<s r( z"rearrange_bins..gsKKK&62KKKr*ci|]\}}|| SrKrK)rLr<new_ids r(rMz"rearrange_bins..is.    F   r*zCreated bin mapping)dictrto_numpyrbf assign_viewdropnagroupbyapply sort_bedframerr1rrrrrupdateindex) bins_oldrF new_chrom_colr! chromdict bins_subset bins_invertedbins_new bin_mappingrDs @r(rearrange_binsra/sb<S1133W]5K5T5T5V5VWWXXI GFO $ $ & &)AS)H(R(R(T(TUUH##8*#55H.7EJJJQQRK Me<< AAAA B B $   !H !/33I>>HW!%8G+<,E,Ec,J,JKKKK  "%hx&8&?&?&D&Dhn"U"U     L&'''}}hZ}00H [  r*逖wc |} t|gd|dd}n"#t$r} td| d} ~ wwxYw| |jd}|ud}||jvrjt jd ||jvrMd tj d tj tj j }||jvM||jvr|d |jdd|f<|ud }||jvrt jd||jvrMdtj d tj tj j }||jvM||jvrd|jdd|f<tj||dstd|dd} t'| |||\} } t jdt)j|| t-|| |||t/dt jd|dS)aReorder cooler following a genomic view. Parameters ---------- clr : cooler.Cooler Cooler object view_df : viewframe Viewframe with new order of genomic regions. Can have an additional column for the new chromosome name (`new_chrom_col`), and an additional column for the strand orientation (`orientation_col`, '-' will invert the region). out_cooler : str File path to save the reordered data new_chrom_col : str, optional Column name in the view_df specifying new chromosome name for each region, by default 'new_chrom'. If None, then the default chrom column names will be used. orientation_col : str, optional Column name in the view_df specifying strand orientation of each region, by default 'strand'. The values in this column can be "+" or "-". If None, then all will be assumed "+". assembly : str, optional The name of the assembly for the new cooler. If None, uses the same as in the original cooler. chunksize : int, optional The number of pixels loaded and processed per step of computation. mode : str, optional (w)rite or (a)ppend to the output cooler file. Default: w )rHr.r-r:FT) check_sorting raise_errorsz0view_df is not a valid viewframe or incompatibleNzgenome-assemblyr8zknew_chrom_col is not provided, but new_chrom column exists. Pre-existing new_chrom column will not be used. new_chrom_rrHrzgorientation_col is not provided, but strand column exists. Pre-existing strand column will not be used.strand_+c`tjtj|jdkS)Nr)ralldiffrY)rCs r(rEz"rearrange_cooler..s"rvbgag>N>NRS>S7T7Tr*z#New chromosomes are not consecutive)r[r!zCreating a new cooler)r geA)assemblymodemergebufzCreated a new cooler at )copyr Exception ValueErrorrrIrwarnrrandomrandintiinfoint32r4locrkrUrVbinsracooler create_coolerr)r) rrF out_coolerr[r!rmr rn_erZr_r`s r(rearrange_coolerrssLllnnGT # 555 6      TTTKLLRSST8-.# GO + + LC    7?22O!2!21bhrx6H6H6L!M!MOO 7?22GO++(/(8 AAA}$%" go - - L@   00V (9(9!RXbh=O=O=S(T(TVVO00go--*- AAA&' 6 &&,,-T-TUU  @>???xxzz!!!}H*'Hk L())) !      S     L8J8899999s2 AA  A)rrr)r8r)r8rNrbrc)bioframerRrznumpyr lib.checksrr basicConfigINFOr)r1r7rarrKr*r(rs 000000',''''KL@@@@, CKA!A!A!A!P  f:f:f:f:f:f:r*