DUfe*ddlmZddlZddlZddlZddlZddlZ ddl m Z ddl m Z ddlmZmZmZmZmZmZmZmZddlmZddlmZmZmZmZd d gZhd ZGd d Z d6dZ! d7d8dZ" d7d8dZ# d9d:dZ$ d;dd5Z&dS)?) annotationsN)is_integer_dtype) coo_matrix) CSRReaderDirectRangeQuery2DFillLowerRangeQuery2DRangeSelector1DRangeSelector2Dgetregion_to_extentregion_to_offset) list_coolers) closing_hdf5 open_hdf5parse_cooler_uri parse_regionCoolerannotate>KRVCVC_SQRTceZdZdZd9d:dZd;d Zdd?dZe d@dZ e dAdZ e dBdZ e dCdZ dDdZdEd!Ze dFd"Ze dGd#ZdHd%ZdHd&ZdIdJd*Z dKdLd7Zd@d8ZdS)Mra] A convenient interface to a cooler data collection. Parameters ---------- store : str, :py:class:`h5py.File` or :py:class:`h5py.Group` Path to a cooler file, URI string, or open handle to the root HDF5 group of a cooler data collection. root : str, optional [deprecated] HDF5 Group path to root of cooler group if ``store`` is a file. This option is deprecated. Instead, use a URI string of the form :file:`::`. kwargs : optional Options to be passed to :py:class:`h5py.File()` upon every access. By default, the file is opened with the default driver and mode='r'. Notes ----- If ``store`` is a file path, the file will be opened temporarily in when performing operations. This allows :py:class:`Cooler` objects to be serialized for multiprocess and distributed computations. Metadata is accessible as a dictionary through the :py:attr:`info` property. Table selectors, created using :py:meth:`chroms`, :py:meth:`bins`, and :py:meth:`pixels`, perform range queries over table rows, returning :py:class:`pd.DataFrame` and :py:class:`pd.Series`. A matrix selector, created using :py:meth:`matrix`, performs 2D matrix range queries, returning :py:class:`numpy.ndarray` or :py:class:`scipy.sparse.coo_matrix`. Nstorestr | h5py.Grouproot str | Nonec <t|tr|t|\|_|_nat j|r>t|fi|5}|jj|_||_dddn #1swxYwYntd|jdz|jz|_ |j|_ ||_ nG|jj|_|j |_|jdz|jz|_ |j|_ i|_ |dS)Nz!Not a valid path to a Cooler file::) isinstancestrrfilenamerh5pyis_hdf5rfile ValueErroruriropen_kwsname_refresh)selfrrkwargsh5s G/var/www/html/software/conda/lib/python3.11/site-packages/cooler/api.py__init__zCooler.__init__DsE eS ! ! |+;E+B+B( tyye$$ Fu////%2$&G$4DM $DI%%%%%%%%%%%%%%%!!DEEE}t+di7DHDJ"DMM"J/DM DI}t+di7DHDJDM sA;;A?A?returnNonec  t|jfi|j5}||j}t |}|dt |d<|dd|_tt|dtt||_ t||_|jdd}|dk|_ddddS#1swxYwYdS#t$$rKd|jd}t'|j}t|r |d|dd zz }t%|dwxYw) Nr)length storage-modesymmetric-upperzNo cooler found at: .z Coolers found in z. z Use '::' to specify a group path)rrr(rchromsastypeobject set_index _chromsizesdictziprangelen _chromidsinfo_infor _is_symm_upperKeyErrorr)r+r-grp_ctmodeerr_msglistings r.r*zCooler._refreshZs .4:7777 @2mSkk!&k0088F #&==#8#8#B !%c#f+uSXX&G&G!H!H!#YY z~~n6GHH&*.?&?# @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ . . .:TZ:::G"4:..G7|| 4444897## - .s0DCD5 DDD D DAE$pathr! np.ndarrayct|jfi|j5}||j}||ddcdddS#1swxYwYdSN)rrr(rr+rJr-rEs r. _load_dsetzCooler._load_dsetos tz 3 3T] 3 3 rTY-Ct9QQQ<                  sAA Ar<ct|jfi|j5}||j}t ||jcdddS#1swxYwYdSrM)rrr(rr<attrsrNs r. _load_attrszCooler._load_attrsts tz 3 3T] 3 3 )rTY-CD (( ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) )s'A  AArrG h5py.Groupc ftj|j|fi||j}t |S)aOpen the HDF5 group containing the Cooler with :py:mod:`h5py` Functions as a context manager. Any ``open_kws`` passed during construction are ignored. Parameters ---------- mode : str, optional [default: 'r'] * ``'r'`` (readonly) * ``'r+'`` or ``'a'`` (read/write) Notes ----- For other parameters, see :py:class:`h5py.File`. )r#Filer"rr)r+rGr,rEs r.openz Cooler.openys4"i t66v66tyAC   c8|jddS)zIndicates whether ordinary sparse matrix encoding is used (``"square"``) or whether a symmetric matrix is encoded by storing only the upper triangular elements (``"symmetric-upper"``). r4r5)rBr r+s r. storage_modezCooler.storage_modes z~~n.?@@@rX int | Nonec|jdS)z-Resolution in base pairs if uniform else Nonezbin-sizerBrZs r.binsizezCooler.binsizesz*%%rX pd.Seriesc|jS)z=Ordered mapping of reference sequences to their lengths in bp)r;rZs r. chromsizeszCooler.chromsizess rX list[str]c4t|jjS)z List of reference sequence names)listr;indexrZs r. chromnameszCooler.chromnamessD$*+++rXregionstr | tuple[str, int, int]intc t|jfi|j5}||j}t ||jt ||j|jcdddS#1swxYwYdS)a'Bin ID containing the left end of a genomic region Parameters ---------- region : str or tuple Genomic range Returns ------- int Examples -------- >>> c.offset('chr3') # doctest: +SKIP 1311 N) rrr(rrr@rr;r_r+rhr-rEs r.offsetz Cooler.offset$tz 3 3T] 3 3 rTY-C#VT%566                   >> c.extent('chr3') # doctest: +SKIP (1311, 2131) N rrr(rr r@rr;r_rls r.extentz Cooler.extentrnroct|jfi|j5}||j}t |cdddS#1swxYwYdS)zUFile information and metadata Returns ------- dict N)rrr(rrA)r+r-rEs r.rAz Cooler.infostz 3 3T] 3 3 rTY-C99                  sAAAc$|jdfdzS)Nnbinsr^rZs r.shapez Cooler.shapes 7#%))rXr c Lfd}td|djdS)z[Chromosome table selector Returns ------- Table selector ctjfij5}|j}t ||||ficdddS#1swxYwYdSrM)rrr(rr7fieldslohir-rEr,r+s r._slicezCooler.chroms.._slices4:7777 =2mc2r6<._slices4:7777 ;2mCR::6:: ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ;rc tjfij5}|j}t |jt |jjcdddS#1swxYwYdSrMrr)rhr-rEr+s r._fetchzCooler.bins.._fetch s4:7777 2m'N )9::L                   s._slice$s4:7777 C2mc2r64BB6BB C C C C C C C C C C C C C C C C C CsAA Ac <tjfij5}|j}t |jt |jj\}}|dd|}|dd|}||fcdddS#1swxYwYdS)Nindexes bin1_offsetrr)rhr-rEi0i1r}r~r+s r.rzCooler.pixels.._fetch)s4:7777 2m)N )9::L B ^M226^M2262v                  sA+BBBNnnzr)r+rr,rrs``` r.rz Cooler.pixelsse C C C C C C C     tVVTZ5FGGGrXT逖fieldbalance bool | strsparse as_pixels ignore_indexdivisive_weights bool | None chunksizer c tvrdfd} dfd } t|| | jdfdzS)aContact matrix selector Parameters ---------- field : str, optional Which column of the pixel table to fill the matrix with. By default, the 'count' column is used. balance : bool, optional Whether to apply pre-calculated matrix balancing weights to the selection. Default is True and uses a column named 'weight'. Alternatively, pass the name of the bin table column containing the desired balancing weights. Set to False to return untransformed counts. sparse: bool, optional Return a scipy.sparse.coo_matrix instead of a dense 2D numpy array. as_pixels: bool, optional Return a DataFrame of the corresponding rows from the pixel table instead of a rectangular sparse matrix. False by default. join : bool, optional If requesting pixels, specifies whether to expand the bin ID columns into (chrom, start, end). Has no effect when requesting a rectangular matrix. Default is True. ignore_index : bool, optional If requesting pixels, don't populate the index column with the pixel IDs to improve performance. Default is True. divisive_weights : bool, optional Force balancing weights to be interpreted as divisive (True) or multiplicative (False). Weights are always assumed to be multiplicative by default unless named KR, VC or SQRT_VC, in which case they are assumed to be divisive by default. Returns ------- Matrix selector Notes ----- If ``as_pixels=True``, only data explicitly stored in the pixel table will be returned: if the cooler's storage mode is symmetric-upper, lower triangular elements will not be generated. If ``as_pixels=False``, those missing non-zero elements will automatically be filled in. NTct jfi j5}| j}t |||||| jcdddS#1swxYwYdSrM)rrr(rmatrixrC)rrrj0j1r-rErrrrrrr+rs r.rzCooler.matrix.._slicers4:7777 2m $'                  s.AAAcdt jfi j5}| j}||}t | j}t | j}t | j| j\}}t | j| j\}}||||fcdddS#1swxYwYdSrM) rrr(rrr;r r@r_) rhregion2r-rEregion1rrrrr+s r.rzCooler.matrix.._fetchs4:7777 &2m?$G&vt/?@@&w0@AA)#t~w UUB)#t~w UUB2r2~ & & & & & & & & & & & & & & & & & &sA?B%%B),B)rvrwrM)_4DN_DIVISIVE_WEIGHTSr rB) r+rrrrrrrrrrs ` ``````` r.rz Cooler.matrix8sn + + +0@0H#             ( & & & & & &ufftz'7J6Lq6PQQQrXct|jtr1tj|j}|d|j}nt|j}d|dS)Nrz )r rr!osrJbasenamerrepr)r+r" containers r.__repr__zCooler.__repr__sd dj# & & )w'' 33H#22ty22IITZ((I(9((((rXrM)rrrr)r0r1)rJr!r0rK)rJr!r0r<)rS)rGr!r0rT)r0r!)r0r\)r0r`)r0rc)rhrir0rj)rhrir0rp)r0r<)r0rp)r0r F)rrr0r )NTFFFTNr)rrrrrrrrrrrrrrrrjr0r )__name__ __module__ __qualname____doc__r/r*rOrRrWpropertyr[r_rbrgrmrsrArxr7rrrrrXr.rr s!!F,....*    )))) !!!!!(AAAXA&&&X&   X ,,,X,66   X ***X*JJJJ JJJJ4!H!H!H!H!HJ!"!(,!YRYRYRYRYRv))))))rXr-rTr0r<ci}|jD]E\}}t|tr& t j|}n#t $rYnwxYw|||<F|S)z File and user metadata dict. Parameters ---------- h5 : :py:class:`h5py.File` or :py:class:`h5py.Group` Open handle to cooler file. Returns ------- dict )rQitemsr r!jsonloadsr&)r-dkvs r.rArAs~ A  1 a    JqMM    ! HsA  AAr}rjr~r\r|list[str] | None pd.DataFramec |etjddgtj|d}t |d|||fi|S)a Table describing the chromosomes/scaffolds/contigs used. They appear in the same order they occur in the heatmap. Parameters ---------- h5 : :py:class:`h5py.File` or :py:class:`h5py.Group` Open handle to cooler file. lo, hi : int, optional Range of rows to select from the table. fields : sequence of str, optional Subset of columns to select from table. Returns ------- :py:class:`DataFrame` Nr)r3r7)pdIndexappendkeysdrop_duplicatesr )r-r}r~r|r,s r.r7r7st2~ Hfh' ( ( VBHR\..0011 2 2 _    r(|RV 6 6v 6 66rXc p|etjgdtj|d}t |d|||fi|}d|vr|dd}t |tr|}n|d}t|j rj|rht|d}tj ||d }t |trtj ||j}n||d<|S) a Table describing the genomic bins that make up the axes of the heatmap. Parameters ---------- h5 : :py:class:`h5py.File` or :py:class:`h5py.Group` Open handle to cooler file. lo, hi : int, optional Range of rows to select from the table. fields : sequence of str, optional Subset of columns to select from table. Returns ------- :py:class:`DataFrame` Nchromstartendrr convert_enumTr))r|)ordered)rrrrrr r r!rdtyper7 Categorical from_codesSeriesrf) r-r}r~r|r,outr chrom_colrgs r.rrs20~ H... / / VBHRZ__..// 0 0 _    bj"b& 3 3F 3 3C&zz.$77 fc " " %IIG I IO , , ) )6222J11)ZQU1VVI&#&& )i 3955(G JrXTrrc X|etjddgtj|d}t |d|||fi|}|r*t |dddgdfi|}t ||d }|S) aG Table describing the nonzero upper triangular pixels of the Hi-C contact heatmap. Parameters ---------- h5 : :py:class:`h5py.File` or :py:class:`h5py.Group` Open handle to cooler file. lo, hi : int, optional Range of rows to select from the table. fields : sequence of str, optional Subset of columns to select from table. join : bool, optional Whether or not to expand bin ID columns to their full bin description (chrom, start, end). Default is True. Returns ------- :py:class:`DataFrame` Nbin1_idbin2_idrrrrTreplace)rrrrrr r)r-r}r~r|rr,dfrs r.rrs:~ Hi+ , , VBHR\..0011 2 2 _    R\2r6 4 4V 4 4B .2f:q$(A(A(ALLVLL b$ - - - IrXFrrpd.DataFrame | RangeSelector1Drc |j t|trd}nd}g}d vr|dt jdd}t|dkrdx}}nNt|t|kr)|| }}nd\}}||||}| |j ||z  d  d d vr|d t jdd} t| dkrdx}}nNt|t|kr)| | }}nd\}}||||} | | j | |z  d  d |r% fddD} || d}tjg|| d d} |j| _| S)a Add bin annotations to a data frame of pixels. This is done by performing a relational "join" against the bin IDs of a table that describes properties of the genomic bins. New columns will be appended on the left of the output data frame. .. versionchanged:: 0.8.0 The default value of ``replace`` changed to False. Parameters ---------- pixels : :py:class:`DataFrame` A data frame containing columns named ``bin1_id`` and/or ``bin2_id``. If columns ``bin1_id`` and ``bin2_id`` are both present in ``pixels``, the adjoined columns will be suffixed with '1' and '2' accordingly. bins : :py:class:`DataFrame` or DataFrame selector Data structure that contains a full description of the genomic bins of the contact matrix, where the index corresponds to bin IDs. replace : bool, optional Remove the original ``bin1_id`` and ``bin2_id`` columns from the output. Default is False. Returns ------- :py:class:`DataFrame` c$||||dzndS)Nrr)selbegrs r. _loc_slicezannotate.._loc_sliceassS1WWTAB BrXc |j||SrM)loc)rrrs r.rzannotate.._loc_slicefs6#c'? "rXrFsafe)copycastingr)rNc |dzS)N1rxs r.zannotate..{ a#grX)columnsT)droprc |dzS)N2rrs r.rzannotate..rrXcg|]}|v| Srr).0colrs r. zannotate..sPPPrX)rrr)axis)rr r to_numpyr8npint64r?minmaxrilocrename reset_indexrrconcatrf)rrrrannsbin1bminbmaxann1bin2ann2 cols_to_droprrs @r.rr<s@nG$(( # C C C C  # # # DGi ))++2228%QW2XX t99>>OD44 YYV $ $TXXZZ$DD JD$z$d++ IdTk " V--V . . [d[ # #   Gi ))++2228%QW2XX t99>>OD44 YYV $ $TXXZZ$DD JD$z$d++ IdTk " V--V . . [d[ # #   3PPPP'=PPP \22 ):d:F..D.99: C C CC CI JrXrrrrrrrrrrrrrr fill_lower&np.ndarray | coo_matrix | pd.DataFramecJ|d}t|tr|}n|rd}|r#||dvrtd|ddzdzt|d |d dd}|rt ||||||f| | }|}|rt ||g}t||d }| r(d||dzz ||dz<d||dzz ||dz<||dz||dzz||z|d<| r;t |gd}t||d }|S|r| rt||||||f| }nt ||||||f| }| }|rb|d|}|||}||f||fkr|n |||}| r d|z }d|z }||j ||j z|j z|_ |S| rt||||||f| }nt ||||||f| }|}|rR|d|}|||}||f||fkr|n |||}| r d|z }d|z }|tj||z}|S)a Two-dimensional range query on the Hi-C contact heatmap. Depending on the options, returns either a 2D NumPy array, a rectangular sparse ``coo_matrix``, or a data frame of pixels. Parameters ---------- h5 : :py:class:`h5py.File` or :py:class:`h5py.Group` Open handle to cooler file. i0, i1 : int, optional Bin range along the 0th (row) axis of the heatmap. j0, j1 : int, optional Bin range along the 1st (col) axis of the heatmap. field : str, optional Which column of the pixel table to fill the matrix with. By default, the 'count' column is used. balance : bool, optional Whether to apply pre-calculated matrix balancing weights to the selection. Default is True and uses a column named 'weight'. Alternatively, pass the name of the bin table column containing the desired balancing weights. Set to False to return untransformed counts. sparse: bool, optional Return a scipy.sparse.coo_matrix instead of a dense 2D numpy array. as_pixels: bool, optional Return a DataFrame of the corresponding rows from the pixel table instead of a rectangular sparse matrix. False by default. join : bool, optional If requesting pixels, specifies whether to expand the bin ID columns into (chrom, start, end). Has no effect when requesting a rectangular matrix. Default is True. ignore_index : bool, optional If requesting pixels, don't populate the index column with the pixel IDs to improve performance. Default is True. Returns ------- ndarray, coo_matrix or DataFrame Notes ----- If ``as_pixels=True``, only data explicitly stored in the pixel table will be returned: if the cooler's storage mode is symmetric-upper, lower triangular elements will not be generated. If ``as_pixels=False``, those missing non-zero elements will automatically be filled in. NcountweightrzNo column 'bins/'z(found. Use ``cooler.balance_cooler`` to z1calculate balancing weights or set balance=False.rzindexes/bin1_offset) return_indexFrrrrbalancedrT)r r!r&rrto_framerrrr to_sparse_matrixrowrdatato_arrayrouter)r-rrrrrrrrrrrrrr)readerenginerweightsdf2rmatbias1bias2arrs r.rrst| }'3  4r&z)) &t & & &8 9A B   r(|R(=%>qqq%A B BF8$ EBB+YEU   __    LRjjoo''/G2w666C 6"#c$*o"5D3J"#c$*o"5D3J _s4#:>UKBzN  2"::??$$%>%>%>?D"dD111B "  T*652r2r:JIVVFF'BB7GSSF%%''  Bj&GBrENE "X"b11EEwr"u~E "E E SW~cg6ACH   T*652r2r:JIVVFF'BB7GSSFoo  /j&GBrENE "X"b11EEwr"u~E "E E ...C rX)r-rTr0r<)rNN) r-rTr}rjr~r\r|rr0r)rNNT) r-rTr}rjr~r\r|rrrr0rr)rrrrrrr0r) NTFFTTFrT)r-rTrrjrrjrrjrrjrrrrrrrrrrrrrrrrjrrr0r)' __future__rrr#numpyrpandasr simplejsonrpandas.api.typesr scipy.sparsercorerrr r r r r rfileopsrutilrrrr__all__rrrAr7rrrrrrXr.r sG"""""" ------######                    "!!!!!IIIIIIIIIIII Z 0//y)y)y)y)y)y)y)y)x     6# 77777H# 55555t# *****`[[[[[H"GGGGGGGrX