tf2bddlmZddlZddlmZddlmZddlmZddl 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'y)%) annotationsN)partial)add)Literal) get_logger) MapFunctor)Cooler) partitionsplit)madbalance_coolerc eZdZy)ConvergenceWarningN)__name__ __module__ __qualname__X/var/www/html/software/conda/envs/higlass/lib/python3.12/site-packages/cooler/balance.pyrrsrrc8tj|ddS)Npixelscount)npcopy)chunks r_initrs 775?7+ ,,rcd||dk7<|S)Nrrr)rdatas r _binarizer sDO Krc\|d}tj|d|dz |k}d||<|S)Nrbin1_idbin2_idr)rabs)n_diagsrrrmasks r _zero_diagsr'"s; 8_F 66&#fY&77 87 BDDJ KrcL|dd}|d}||d||dk7}d||<|SNbinschromrr"r#rrrr chrom_idsrr&s r _zero_transr.)Ef g&I 8_F VI& '9VI5F+G GDDJ KrcL|dd}|d}||d||dk(}d||<|Sr)rr,s r _zero_cisr11r/rc8|d}||d||dz|z}|S)Nrr"r#r)vecrrrs r_timesouterproductr49s3 8_F vi !Cy(9$: :T AD Krct|dd}|d}tj|d||tj|d||z}|S)Nr*r+rr")weights minlengthr#)lenrbincount)rrnrmargs r _marginalizer<?sY E&M' "#A 8_F ;;vi($! Dr{{y41H D Krc d} t|} t|D] } t|||| jtj |j t |j tjttj| } | | dk7}t|s'tj} tj|ddd}nl| |jz } d| | dk(<|| z}|j}tj!d|||ks nt#j$dt&j} tj||dk(<|r|tj(| z}|| fS)N?spansmapuse_lockrr variance is ,Iteration limit reached without convergence.)r8ranger preparerpiper4r<reducerrzerosnanmeanvarloggerinfowarningswarnrsqrt)biasclrr@filters chunksizerAtol max_itersrescale_marginalsrBscalen_bins_r;nzmargrMs r_balance_genomewider^HsM E YF 9   #Uh ? WU^ T'] T$d + T,  VC&) * dai6{FFEffDGC fkkm#TQY  jjl l3%() 9 1 4  :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}srZZ\& !! $F #cjjl+,INN#9:M>>"78L WWS^ $F VRVV,I YF9mCR&8-:KL.* R F3K #\"%5SYsC34ffy! AcC(Cg($/l#RXXf-. 2;D$!)_Fv; ffR &++-'DDO BK4 K**,C KK,se, -Sy3 8 MMA&+aP"    BKFF!q& s  #  BK2775> )K].*`  ""rc d} t|} |jd} dtjt | dd| ddD cgc]\} }d|| z | z z g|| z zc}} z }t |D]!}t |||| jtj|jtjt||zjtjttj| }||dk7}t|s'tj } tj |ddd}nl||j#z }d||dk(<||z}|j%}t&j)d|||ks"nt+j,d t.j#} tj ||dk(<|r|tj0| z}|| fScc}} w) Nr>rarbrr?rrCrDrE)r8rfr concatenaterirFr rGrrHr1r4r<rIrrJrKrLrMrNrOrPrQrrR)rSrTr@rUrVrArWrXrYrBrZr[rkrprqcweightsr\r;r]rMs r_balance_transonlyrys E YFNN#9:MR^^mCR0-2CD B27f$$ &"r' 2 H9   #Uh ? WU^ T'] T)_ T$dXo 6 T,  VC&) * dai6{FFEffDGC fkkm#TQY  jjl l3%() 9 3 6  :iweight)cis_only trans_only ignore_diagsmad_maxmin_nnz min_count blacklistrYx0rWrXrVrArBstore store_namec t|jd}| |} d|fg}n5tjd|| z| }t t |dd|dd}g}|r|j t|r|j tt|t|jd}| | }d|tj|<ntj|t}|dkDr|tg|}t||| |jt j#|j#t$j't(tj*|}d|||k<|}t||| |jt j#|j#t$j't(tj*|}|rd|||k<|dkDr|j-d }t |dd|ddD]1\}}|||}|||xxxtj.||dkDzccc3tj0||dkD}tj.|}t3|} tj4||| zz }!d|||!k<|d||<|rt7||||| | | | || \}}"}#n3|rt9||||| | | | || \}}"}#nt;||||| | | | || \}}"}#| ||||||"|#| k|#d d }$|ri|j=d 5}%||%d vr|%d |=ddd}&|%d j>|fd|i|&|%d |j@jC|$ddd||$fS||$fS#1swY||$fSxYw)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)"intrOrrerjriappendr.rr'isnanrgfloatr r rGrrHr<rIrrJrfmedianlogr expruryr^opencreate_datasetattrsupdate)'rTrrrrrrrrYrrWrXrVrArBrrrr@edges base_filtersr[rSrUmarg_nnzr;offsetsrprqc_marg logNzMarg med_logNzMarg dev_logNzMargcutoffrZrMstatsgrph5optss' rrrsp chhuo C S  !S9_i8SsU12Y/0LK(GK>?'" #F ~ RXXd^wwvU+{,|, #Uh ? WU^ T'] T,  VC&) * $%X  G cC(;  g l  RXXf% &  !"TI {..!78'#2, 4 9FB"R[F BK299VFQJ%78 8K 9FF4q>*  ), I  -(??@TF]Y+            eS -            eS/            eS$3Y! E XXd^ 8sS[(K +%+CF &CK & &z G G G K # ) ) 0 0 7  8 ;4; 8 ;s A M((M4)$rTr rboolrrrzint | Literal[False]rrrrrrrz str | NonerYrrznp.ndarray | NonerWrrXrrVrrAr rBrrrrstrreturnztuple[np.ndarray, dict])( __future__rrP functoolsroperatorrtypingrnumpyr_loggingr_typingr apir parallelr r utilr __all__rrN UserWarningrrr r'r.r1r4r<r^ruryrAriterative_correctionrrrrs" &   H   -  2jD#N;B)* " %V VV V ' V  VVVVV V VVV V !V" #V$%V&'Vr&r