DUf2*ddlmZddlZddlmZddlmZddlmZddl Z ddl m Z ddl mZdd lmZdd lmZmZdd lmZd gZe eZGd deZdZdZdZdZdZdZ dZ!dZ"dZ#dZ$dddddddddddd e%ddd!d"d@d?Z&e&Z'dS)A) annotationsN)partial)add)Literal) get_logger) MapFunctor)Cooler) partitionsplit)madbalance_coolerceZdZdS)ConvergenceWarningN)__name__ __module__ __qualname__K/var/www/html/software/conda/lib/python3.11/site-packages/cooler/balance.pyrrsDrrcBtj|ddS)Npixelscount)npcopy)chunks r_initrs 75?7+ , ,,rcd||dk<|S)Nrrr)rdatas r _binarizer sDO Krcn|d}tj|d|dz |k}d||<|S)Nrbin1_idbin2_idr)rabs)n_diagsrrrmasks r _zero_diagsr'"s< 8_F 6&#fY&77 8 87 BDDJ Krcx|dd}|d}||d||dk}d||<|SNbinschromrr"r#rrrr chrom_idsrr&s r _zero_transr.)Ef g&I 8_F VI& '9VI5F+G GDDJ Krcx|dd}|d}||d||dk}d||<|Sr)rr,s r _zero_cisr11r/rcV|d}||d||dz|z}|S)Nrr"r#r)vecrrrs r_timesouterproductr49s3 8_F vi !Cy(9$: :T AD Krct|dd}|d}tj|d||tj|d||z}|S)Nr*r+rr")weights minlengthr#)lenrbincount)rrnrmargs r _marginalizer<?sj E&M' "##A 8_F ;vi($! D D Dr{y41HHH D Krc rd} t|} t|D]7} t|||| t|t |tttj | } | | dk}t|s!tj } tj |dd<d}nz| | z } d| | dk<|| z}|}td|||krn9t#jdt&| } tj ||dk<|r|tj| z}|| |fS)N?spansmapuse_lockrr variance is ,Iteration limit reached without convergence.)r8ranger preparerpiper4r<reducerrzerosnanmeanvarloggerinfowarningswarnrsqrt)biasclrr@filters chunksizerAtol max_itersrescale_marginalsrBscalen_bins_r;nzmargrMs r_balance_genomewider^Hs E YYF 9     #Uh ? ? ? WU^^ T']] T$d + + T,   VC&)) * * dai6{{ FEfDGC Efkkmm#TQY  jjll (3(())) 99 E   :r?rrCrDz/Iteration limit reached without convergence on .)chromsraranger8 _load_dsetones full_likerKziprNrOlistr rFr rGrrHr4r<rIrrJrLrMrPrQrrR)rSrTr@rUrVrArWrXrYrBrdr- chrom_offsets bin1_offsetsscales variancesr[cidlohiplophirZrMr\r;r]bs r_balance_cisonlyru}sZZ\\& !!!! $F #cjjll++,,INN#9::M>>"788L WS^^ $ $F VRV,,I YYF9mCRC&8-:KLL.*.* R F3K   #\"%5SYsC3344fy!!  AcC(CCCg($//l##RXf--.. 2;D$!)_Fv;;  fRU &++--'DDO BKKK4 KKK**,,C KK,s,, - - -Syy MP&+PPP"      BKF!q& s  #  * BKKK275>> )KKK  ""rc Vd} t||d} dtjfdt | dd| ddDz } t |D]R} t |||| t | t t|| z t ttj}||dk}t|s!tj} tj|dd<d}nz||z }d||dk<||z}|}t&d |||krnTt+jd t.|} tj||dk<|r|tj| z}|| |fS) Nr>rac:g|]\}}d||z z z g||z zS)rr).0rprqr[s r z&_balance_transonly..sG   B27f$$ &"r' 2   rrbrr?rrCrDrE)r8rfr concatenaterirFr rGrrHr1r4r<rIrrJrKrLrMrNrOrPrQrrR)rSrTr@rUrVrArWrXrYrBrZrkcweightsr\r;r]rMr[s @r_balance_transonlyr|s E YYFNN#9::MR^    mCRC0-2CDD   H9     #Uh ? ? ? WU^^ T']] T)__ T$dXo 6 6 T,   VC&)) * * dai6{{ FEfDGC Efkkmm#TQY  jjll (3(())) 99 E   :iweight)cis_only trans_only ignore_diagsmad_maxmin_nnz min_count blacklistrYx0rWrXrVrArBstore store_namerTr rboolrrint | Literal[False]rintrrr str | NonerYrnp.ndarray | NonerWfloatrXrVrAr rBrrstrreturntuple[np.ndarray, dict]c t|jd}| |} d|fg}nFtjd|| z| }t t |dd|dd}g}|r|t|r(|tt|t|jd}| | }d|tj |<ntj |t}|dkrtg|}t||| |t |t$t(tj|}d|||k<|}t||| |t |t$t(tj|}|r d|||k<|dkr|d }t |dd|ddD]=\}}|||}|||xxtj||dkzcc<>tj||dk}tj|}t3|} tj||| zz }!d|||!k<|d||<|rt7||||| | | | || \}}"}#n;|rt9||||| | | | || \}}"}#nt;||||| | | | || \}}"}#| ||||||"|#| k|#d d }$|r|d 5}%||%d vr |%d |=ddd}&|%d j|fd|i|&|%d |j !|$dddn #1swxYwY||$fS)a Iterative correction or matrix balancing of a sparse Hi-C contact map in Cooler HDF5 format. Parameters ---------- clr : cooler.Cooler Cooler object cis_only : bool, optional Do iterative correction on intra-chromosomal data only. Inter-chromosomal data is ignored. trans_only : bool, optional Do iterative correction on inter-chromosomal data only. Intra-chromosomal data is ignored. ignore_diags : int or False, optional Drop elements occurring on the first ``ignore_diags`` diagonals of the matrix (including the main diagonal). chunksize : int or None, optional Split the contact matrix pixel records into equally sized chunks to save memory and/or parallelize. Set to ``None`` to use all the pixels at once. mad_max : int, optional Pre-processing bin-level filter. Drop bins whose log marginal sum is less than ``mad_max`` median absolute deviations below the median log marginal sum. min_nnz : int, optional Pre-processing bin-level filter. Drop bins with fewer nonzero elements than this value. min_count : int, optional Pre-processing bin-level filter. Drop bins with lower marginal sum than this value. blacklist : list or 1D array, optional An explicit list of IDs of bad bins to filter out when performing balancing. rescale_marginals : bool, optional Normalize the balancing weights such that the balanced matrix has rows / columns that sum to 1.0. The scale factor is stored in the ``stats`` output dictionary. map : callable, optional Map function to dispatch the matrix chunks to workers. Default is the builtin ``map``, but alternatives include parallel map implementations from a multiprocessing pool. x0 : 1D array, optional Initial weight vector to use. Default is to start with ones(n_bins). tol : float, optional Convergence criterion is the variance of the marginal (row/col) sum vector. max_iters : int, optional Iteration limit. store : bool, optional Whether to store the results in the file when finished. Default is False. store_name : str, optional Name of the column of the bin table to save to. Default name is 'weight'. Returns ------- bias : 1D array, whose shape is the number of bins in ``h5``. Vector of bin bias weights to normalize the observed contact map. Dropped bins will be assigned the value NaN. N[i, j] = O[i, j] * bias[i] * bias[j] stats : dict Summary of parameters used to perform balancing and the average magnitude of the corrected matrix's marginal sum at convergence. nnzNrrbrnbins)dtyper?raF) rWrrrrrrZ convergedrMdivisive_weightszr+r*gzip) compressioncompression_optsr)"rrOrrerjriappendr.rr'isnanrgrr r rGrrHr<rIrrJrfmedianlogr exprur|r^opencreate_datasetattrsupdate)'rTrrrrrrrrYrrWrXrVrArBrrrr@edges base_filtersr[rSrUmarg_nnzr;offsetsrprqc_marg logNzMarg med_logNzMarg dev_logNzMargcutoffrZrMstatsgrph5optss' rrrsp chuo  C S  !S9_i88SssU122Y//00L)K(((@GK>>???'" # #F ~ RXd^^wvU+++{{,|, #Uh ? ? ? WU^^ T']] T,   VC&)) * * $%X  G cC(;;;  g l   RXf%% & &  #!"TI {{..!788'#2#, 44 9 9FB"R%[F BKKK29VFQJ%788 8KKKKF4q>**  ),, I  -(??@@TF]Y& +             eSS  -             eSS/             eS$3Y!  E 8 XXd^^ 8sS[((K +%+CCF &CK &z G G G G G G K # ) 0 0 7 7 7  8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 ;sAO33O7:O7)$rTr rrrrrrrrrrrrrrrYrrrrWrrXrrVrrAr rBrrrrrrr)( __future__rrP functoolsroperatorrtypingrnumpyr_loggingr_typingr apir parallelr r utilr __all__rrN UserWarningrrr r'r.r1r4r<r^rur|rAriterative_correctionrrrrs"""""" &&&&&&&&   H          ---  222jD#D#D#N;;;B)* " %VVVVVVr&r