o Nrfe@s(ddlmZddlZddlZddlZddlZddlZ ddl m Z ddl m Z ddlmZmZmZmZmZmZmZmZddlmZddlmZmZmZmZd d gZhd ZGd d d Z d9ddZ!  d:d;ddZ"  d:d;ddZ#   dd?d&d Z%   # #   # ' d@dAd7d8Z&dS)B) annotationsN)is_integer_dtype) coo_matrix) CSRReaderDirectRangeQuery2DFillLowerRangeQuery2DRangeSelector1DRangeSelector2Dgetregion_to_extentregion_to_offset) list_coolers) closing_hdf5 open_hdf5parse_cooler_uri parse_regionCoolerannotate>ZKRZVC_SQRTZVCc@seZdZdZdKdLddZdMd d ZdNddZdOddZdPdQddZe dRddZ e dSddZ e dTd d!Z e dUd#d$Z dVd(d)ZdWd+d,Ze dXd-d.Ze dYd/d0ZdZd2d3ZdZd4d5Zd[d\d9d:Z  ; 6 6 6 ;  ::`. kwargs : optional Options to be passed to :py:class:`h5py.File()` upon every access. By default, the file is opened with the default driver and mode='r'. Notes ----- If ``store`` is a file path, the file will be opened temporarily in when performing operations. This allows :py:class:`Cooler` objects to be serialized for multiprocess and distributed computations. Metadata is accessible as a dictionary through the :py:attr:`info` property. Table selectors, created using :py:meth:`chroms`, :py:meth:`bins`, and :py:meth:`pixels`, perform range queries over table rows, returning :py:class:`pd.DataFrame` and :py:class:`pd.Series`. A matrix selector, created using :py:meth:`matrix`, performs 2D matrix range queries, returning :py:class:`numpy.ndarray` or :py:class:`scipy.sparse.coo_matrix`. Nstorestr | h5py.Grouproot str | NonecKst|trM|durt|\|_|_n*t|r8t|fi|}|jj|_||_Wdn1s2wYnt d|jd|j|_ |j|_ ||_ n|jj|_|j |_|jd|j|_ |j|_ i|_ |dS)Nz!Not a valid path to a Cooler file::) isinstancestrrfilenamerh5pyZis_hdf5rfile ValueErroruriropen_kwsname_refresh)selfrrkwargsh5r'S/var/www/html/software/conda/envs/catlas/lib/python3.10/site-packages/cooler/api.py__init__Ds&     zCooler.__init__returnNonecCszXt|jfi|jC}||j}t|}|dt|d<|dd|_t t |dt t ||_ t||_|jdd}|dk|_WdWdS1sQwYWdSty~d|jd}t|j}t |ry|d|dd 7}t|dw) Nr"length storage-modesymmetric-upperzNo cooler found at: .z Coolers found in z. z Use '::' to specify a group path)rrr!rchromsastypeobjectZ set_index _chromsizesdictziprangelen _chromidsinfo_infor _is_symm_upperKeyErrorr)r$r&grpZ_ctmodeerr_msgZlistingr'r'r(r#Zs,   &    zCooler._refreshpathr np.ndarraycCsRt|jfi|j}||j}||ddWdS1s"wYdSN)rrr!rr$r@r&r=r'r'r( _load_dsetos $zCooler._load_dsetr4cCsPt|jfi|j}||j}t||jWdS1s!wYdSrB)rrr!rr4attrsrCr'r'r( _load_attrsts  $zCooler._load_attrsrr> h5py.GroupcKs$tj|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`. )rFilerrr)r$r>r%r=r'r'r(openysz Cooler.opencCs|jddS)zIndicates whether ordinary sparse matrix encoding is used (``"square"``) or whether a symmetric matrix is encoded by storing only the upper triangular elements (``"symmetric-upper"``). r-r.)r:r r$r'r'r( storage_modeszCooler.storage_mode int | NonecCs |jdS)z-Resolution in base pairs if uniform else Nonezbin-sizer:rKr'r'r(binsizes zCooler.binsize pd.SeriescCs|jS)z=Ordered mapping of reference sequences to their lengths in bp)r3rKr'r'r( chromsizesszCooler.chromsizes list[str]cCs t|jjS)z List of reference sequence names)listr3indexrKr'r'r( chromnamess zCooler.chromnamesregionstr | tuple[str, int, int]intcC\t|jfi|j}||j}t||jt||j|jWdS1s'wYdS)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!rr r8rr3rOr$rVr&r=r'r'r(offset  $z Cooler.offsettuple[int, int]cCrY)aGBin IDs containing the left and right ends of a genomic region Parameters ---------- region : str or tuple Genomic range Returns ------- 2-tuple of ints Examples -------- >>> c.extent('chr3') # doctest: +SKIP (1311, 2131) N rrr!rr r8rr3rOrZr'r'r(extentr\z Cooler.extentcCsJt|jfi|j}||j}t|WdS1swYdS)zUFile information and metadata Returns ------- dict N)rrr!rr9)r$r&r=r'r'r(r9s $z Cooler.infocCs|jdfdS)NnbinsrNrKr'r'r(shapesz Cooler.shaper c s"fdd}td|djdS)z[Chromosome table selector Returns ------- Table selector cXtjfij}|j}t||||fiWdS1s%wYdSrB)rrr!rr0fieldslohir&r=r%r$r'r(_slice $zCooler.chroms.._sliceNZnchromsr r:)r$r%rir'rhr(r0s z Cooler.chromsc s.fdd}fdd}td||jdS)zTBin table selector Returns ------- Table selector crcrB)rrr!rbinsrdrhr'r(rirjzCooler.bins.._slicecs\tjfij}|j}t|jt|jjWdS1s'wYdSrBr^)rVr&r=rKr'r(_fetch s  $zCooler.bins.._fetchNr`rk)r$r%rirmr'rhr(rls  z Cooler.binsFjoinboolc s0fdd}fdd}td||jdS)aPixel table selector Parameters ---------- join : bool, optional Whether to expand bin ID columns into chrom, start, and end columns. Default is ``False``. Returns ------- Table selector csZtjfij}|j}t||||fiWdS1s&wYdSrB)rrr!rpixelsrdrnr%r$r'r(ri$s $zCooler.pixels.._slicecstjfij0}|j}t|jt|jj\}}|dd|}|dd|}||fWdS1s=wYdS)NZindexesZ bin1_offsetr^)rVr&r=i0i1rfrgrKr'r(rm)s  $zCooler.pixels.._fetchNZnnzrk)r$rnr%rirmr'rqr(rps  z Cooler.pixelsT逖fieldbalance bool | strsparse as_pixels ignore_indexdivisive_weights bool | None chunksizer c sVtvr dur dfdd} d fdd } t|| | jdfdS) 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. NTcsftjfij}|j}t||||||jWdS1s,wYdSrB)rrr!rmatrixr;)rurrrsj0j1r&r=ryrvr}r{rzrnr$rxr'r(rirs& $zCooler.matrix.._slicec stjfij;}|j}|dur|}t|j}t|j}t|j|j\}}t|j|j\}}||||fWdS1sHwYdSrB) rrr!rrr3r r8rO) rVZregion2r&r=Zregion1rrrsrrrKr'r(rms    $zCooler.matrix.._fetchr`rarB)_4DN_DIVISIVE_WEIGHTSr r:) r$rurvrxryrnrzr{r}rirmr'rr(r~8s 7 z Cooler.matrixcCsBt|jtrtj|j}|d|j}nt|j}d|dS)Nrz )rrrosr@basenamerrepr)r$r containerr'r'r(__repr__s   zCooler.__repr__rB)rrrr)r*r+)r@rr*rA)r@rr*r4)rG)r>rr*rH)r*r)r*rM)r*rP)r*rR)rVrWr*rX)rVrWr*r])r*r4)r*r])r*r F)rnror*r )NTFFFTNrt)rurrvrwrxroryrornrorzror{r|r}rXr*r )__name__ __module__ __qualname____doc__r)r#rDrFrJpropertyrLrOrQrUr[r_r9rbr0rlrpr~rr'r'r'r(r sD #               % [r&rHr*r4c CsPi}|jD]\}}t|tr!zt|}Wn ty Ynw|||<q|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 )rEitemsrrjsonloadsr)r&dkvr'r'r(r9s   r9rfrXrgrMrelist[str] | None pd.DataFramecKsH|durtddgt|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"r,r0)pdIndexappendkeysdrop_duplicatesr )r&rfrgrer%r'r'r(r0s  r0c Ks|durtgdt|d}t|d|||fi|}d|vrb|dd}t|tr6|}n|d}t|j rb|rbt |dd}tj j ||dd }t|tr^t ||j}|S||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` Nchromstartendrlr convert_enumTr")re)Zordered)rrrrrr rrrZdtyper0Z CategoricalZ from_codesZSeriesrT) r&rfrgrer%outrZ chrom_colrUr'r'r(rls&      rlTrnrocKs||durtddgt|d}t|d|||fi|}|r._loc_slicecSs|j||SrB)loc)rrrr'r'r(resrFsafe)copyZcastingr)rNcS|dS)N1r'xr'r'r(zzannotate..columnsT)droprcSr)N2r'rr'r'r(rrcsg|]}|vr|qSr'r').0colrr'r( szannotate..)rrr)Zaxis)rrr Zto_numpyr1npZint64r7minmaxrZilocrTrenameZ reset_indexrrconcat) rprlrrannsZbin1ZbminbmaxZann1Zbin2Zann2Z cols_to_droprr'rr(r<sR        rtrrrsrrrurrvrwrxryrzr{r} fill_lower&np.ndarray | coo_matrix | pd.DataFramecCs|durd}t|tr|}n|rd}|r&||dvr&td|dddt|d |d dd}|rt||||||f| | d }|}|rt||g}t||d d }| rsd||d||d<d||d||d<||d||d|||d<| rt|gd}t||dd }|S|r| rt ||||||f| }n t||||||f| }| }|r|d|}|||}||f||fkr|n|||}| rd|}d|}||j ||j |j |_ |S| rt ||||||f| }n t||||||f| }|}|rA|d|}|||}||f||fkr(|n|||}| r9d|}d|}|t||}|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. NcountweightrlzNo column 'bins/'z(found. Use ``cooler.balance_cooler`` to z1calculate balancing weights or set balance=False.rpzindexes/bin1_offset)Z return_indexFrrrrZbalancedrT)rrrrrZto_framerrlrrZto_sparse_matrixrowrdataZto_arrayrouter)r&rrrsrrrurvrxryrnrzr{r}rr"readerZenginerweightsZdf2rlmatZbias1Zbias2Zarrr'r'r(r~st>  $     "r~)r&rHr*r4)rNN) r&rHrfrXrgrMrerr*r)rNNT) r&rHrfrXrgrMrerrnror*rr)rprrlrrror*r) NTFFTTFrtT)r&rHrrrXrsrXrrXrrXrurrvrwrxroryrornrorzror{ror}rXrror*r)' __future__rrrnumpyrZpandasrZ simplejsonrZpandas.api.typesrZ scipy.sparsercorerrrr r r r r Zfileopsrutilrrrr__all__rrr9r0rlrprr~r'r'r'r(sT   (  ~ $ : . c