o NrfR@s4ddlZddlZddlmmZddlmmZ ddl Z ddl Z ddl ZddlZddlZeeZiaiZddded<ddded<ddded<ddded<dZd d Zd d Z d)ddZddZddZddZ  d*ddZddZddZ ddZ!dd Z"d!d"Z#d#d$Z$d%d&Zd'd(Z%dS)+NICEweightnamevalueKRVCVC_SQRTcCs`z t||kdd}Wnty|jdYSw||}|||}|||||fS)zGet bin ID from absolute coordinates. Args: c (Cooler): Cooler instance of a .cool file. abs_pos (int): Absolute coordinate to be translated. Returns: int: Bin number. rZnbins)npZ flatnonzero IndexErrorinfooffset)cZabs_poschromschrom_cum_lengths chrom_sizesZchr_idchromZrel_posr]/var/www/html/software/conda/envs/catlas/lib/python3.10/site-packages/clodius/tiles/cooler.pyabs_coord_2_bins   rcCs2|j}t|j}tjdt|jjf}|||fS)z Get the chromosome names and cumulative lengths: Args: c (Cooler): A cooler file Return: (names, sizes, lengths) -> (list(string), dict, np.array(int)) r)Z chromnamesdict chromsizesr Zr_Zcumsumvalues)rZ chrom_namesrrrrr"get_chromosome_names_cumul_lengths/s  rdefaultcCst|}t|\}} } t|||| | } t|||| | } t|||| | } t|||| | }|jddd}| |jdksA| |jdkrad\} } } }tjgddtgggd tgggd ffSt| |jdd} t||jdd}|| | d| |df} gd }|d krd | vs|d kr| d n |d vr| ||j dd|}t ||}| |d|d|d<| |d|d|d<|| | d}|| |d}| |d|d|d<| |d|d|d<| |d|d|d<| |d|d|d<|d krd | vs|d kr4|d|d|d|d<|gd||ffS|d vra|d||d||d|d<|||d <|||d <|gd||ffS|gd d!fS)"a>Get balanced pixel data. Args: f: h5py.File An HDF5 Group that contains the cooler for this resolution start_pos_1 (int): Test. end_pos_1 (int): Test. start_pos_2 (int): Test. end_pos_2 (int): Test. Returns: DataFrame: Annotated cooler pixels. TF)Z as_pixelsZbalancerr )rrrr) genome_start1 genome_start2balanced)columns) genome_start genome_endr)rstartendrr)rrr )Z convert_enumZchrom1start1rZchrom2start2rrr#r!r$r"countZweight1Zweight2r12)rrr')NN) coolerCoolerrrmatrixshapepdZ DataFrameminbinsappendZannotate)fZ start_pos_1Z end_pos_1Z start_pos_2Z end_pos_2Z transform resolutionrrrrZi0i1Zj0Zj1r,Zpixelscolsr0bins1bins2rrrget_dataAsV      "   r8cCsHt|d}|jd}|durtdtdt|d}t |\}}}t |d}|jd}t |t |jd}|t d |} i} t |D]6} |t | d } d | vrad d d | d <d| vrlddd | d<d| vrwddd | d<d| vrddd | d<qLddg||g|| t | d} Wd| S1swY| S)zGet information of a cooler file. Args: file_path (str): Path to a cooler file. Returns: dict: Dictionary containing basic information about the cooler file. rzmax-zoomNz no zoom foundz$The `max_zoom` attribute is missing.0zbin-sizer0rrrrrr g)min_posmax_posmax_zoom max_widthZbins_per_dimension transforms)h5pyFileattrsgetloggerr ValueErrorr*r+rintstr TILE_SIZEranger) file_pathr2r?rrrr total_lengthZbin_sizer@rAi f_for_zoomrrrr_get_info_multi_v1sD      ))rPcs&fddtt|tdDS)Ncsg|]}d|qS)r<r.0xbase_resrr s z(get_zoom_resolutions..r )rKhgutZget_quadtree_depthrJ)rrUrrTrget_zoom_resolutionss rXcCs6tjj|dd}t||}tddd|DdS)ze Print comma-separated list of zoom resolutions for a given genome and base resolution. T)Z all_names,css|]}t|VqdS)N)rI)rRresrrr sz)print_zoom_resolutions..N)r*utilZread_chromsizesrXprintjoin)Zchromsizes_filerUr resolutionsrrrprint_zoom_resolutionss r`r c& Cs d}||}||} |||} ||} |||} t|} t| \}}}t|}t|| | d| | d||d\}\}}i}td|D]<}td|D]3}|||} ||d|} |||} ||d|} ||d| k}||d| k}||d| k}||d| k}|}|dj| |t}|dj| |t}d|vrt |dj}nt |dj}t j d t j d }||||f<|d urv|d urv||d | k}||d | k}||d | k}||d | k}|t |d } |t |d }!| d j| |t}"|!d j| |t}#t t|t| t|| |t}$t t|t| t|| |t}%|$|$dk}$|%|%dk}%t j|d d |"f<t j||#d d f<t j|d d |$f<t j||%d d f<||||||f<qNqF|S)a Generate tiles for a given location. This function retrieves tiles for a rectangular region of width x_width and height y_width Parameters --------- hdf_for_resolution: h5py.File An HDF group containing the cooler for the given resolution x_pos: int The starting x position y_pos: int The starting y position cooler_file: string The filename of the cooler file to get the data from x_width: int The number of tiles to retrieve along the x dimension y_width: int The number of tiles to retrieve along the y dimension Returns ------- data_by_tilepos: {(x_pos, y_pos) : np.array} A dictionary of tile data indexed by tile positions r r )r3rrrrr')r r )ZdtypeNr!r)r*r+rsumrr8rKZastyperHr Z nan_to_numZzerosZfloat32isnanarraynanZravel)&hdf_for_resolutionr3Zx_posZy_postransform_typeZx_widthZy_width BINS_PER_TILEZ tile_sizer%Zend1r&Zend2rrrrrMdatar6r7Zdata_by_tileposZx_offsetZy_offsetdfZbinsizejrNvoutZ sub_bins1Z sub_bins2Z nan_bins1Z nan_bins2ZbiZbjZbend1Zbend2rrr make_tiless!         IrmcCsZtt}|D]#}|d}ttt|dd}|d}t|}|||f|q|S)ap Place these tiles into separate lists according to their zoom level and transform type Parameters ---------- tile_ids: [str,...] A list of tile_ids (e.g. xyx.0.0.1) identifying the tiles to be retrieved Returns ------- tile_lists: {(zoomLevel, transformType): [tile_id, tile_id]} A dictionary of tile ids .r r) col defaultdictsetsplitlistmaprHget_transform_typeadd)tile_idsZ tile_id_liststile_id tile_id_parts tile_position zoom_leveltransform_methodrrr%bin_tiles_by_zoom_level_and_transforms  r~cCs*|d}t|dkr|d}|Sd}|S)z Get the transform type specified in the tile id. Parameters ---------- cooler_tile_id: str A tile id for a 2D tile (cooler) Returns ------- transform_type: str The transform type requested for this tile rnror)rslen)ryrzr}rrrrvs rvcCsZt}|d}d|vr|dd|vr|dd|vr"|dd|vr+|d|S)a  Get the available resolutions from a single cooler file. Parameters ---------- cooler: h5py File A cooler file containing binned 2D data Returns ------- transforms: dict A list of transforms available for this dataset r0rrrr )rrrw)r*rArOrrrget_available_transformss     rcCst|d}d|vrdttttt|di}||gt|<i}|dD]}t |dt |||<q't j | }dd|D|d<t|dd}tt|d|dd}||g|d <d d g|d <t|d|}d d|jD|d <d|jvr|jddkrd|d<||fSt|}t|d}dd|jD|d <dd|d D|d <dd|d D|d <t|d|d<t|d|d<d|vrt|d|d<d|jvr|jdsd|d<d|jvr|jddkrd|d<||gt|<||fS)zJ Create the file handle and tileset info for a cooler tileset r9r_cSsg|]}t|qSr)transform_descriptionsrRtrrrrVszmake_mats..rArrlengthr>r r=cS g|] }|dt|dgqSrr rHrQrrrrV rz storage-modeZsquarefalseZ mirror_tilesr:cSrrrrQrrrrVrcSg|]}t|qSrrrRmrrrrVcSrrrrrrrrVrr?r@Z symmetric)rBrCtuplesortedrurHrtkeysmatsrrIrr intersectionrrar*r+ritemsrrP)filepathr2rZ#available_transforms_per_resolutionr3Zall_available_transformsZ genome_lengthrrrr make_matssN "       rcCs$|tvr t|dSt|\}}|S)z Get the tileset info for a cooler file Parameters: ----------- filepath: str The location of the cooler file r )rr)rr2rrrr tileset_infos  rcCs@|d}|d}|dd}t|}d|g||g}|S)a Add a transform type to a cooler tile id if it's not already present. Parameters ---------- tile_id: str A tile id (e.g. xyz.0.1.0) Returns ------- new_tile_id: str A formatted tile id, potentially with an added transform_type rnrr ro)rsrvr^)ryrzZ tileset_uuidr{rf new_tile_idrrradd_transform_type&s  rc Csji}g}|D]}t|}|||<||g7}qt||}g}|D]\}}||vr2||} || |fg7}q|S) )rgenerate_tiles) rrxZtransform_id_to_original_idZ new_tile_idsryrgenerated_tilesZtiles_to_returnZ tile_valueZoriginal_tile_idrrrtiles>s   rc sd}|tvr t|t|}t|}ttjdd|D}g}|D]}t|ddd|dddt |d|d|d}dvrmt dddDd d } t | kr`q$| } |dt | } nd krtq$|t } d d |} dd|D} tt fdd| } tt fdd| } t | dkrq$tdd| D} tdd| D}tdd| D}tdd| D}t| | | ||| d||d}fdd|D}||7}q$|S)a Generate tiles from a cooler file. Parameters ---------- tileset: tilesets.models.Tileset object The tileset that the tile ids should be retrieved from tile_ids: [str,...] A list of tile_ids (e.g. xyx.0.0.1) identifying the tiles to be retrieved Returns ------- generated_tiles: [(tile_id, tile_data),...] A list of tile_id, tile_data tuples r cSsg|]}t|qSr)rWZpartition_by_adjacent_tilesrrrrrVqsz"generate_tiles..rrnr r_cSrrr)rRr9rrrrVrT)reverser?r@r<cSs(g|]}dd|dddDqS)cSrrrrQrrrrVrz-generate_tiles...rnr<ro)rsrrrrrVs(cs|ddddkS)Nrr>r rrSrrrz generate_tiles..cs|ddddkS)Nr r>rrrrrrrcSg|]}|dqSrrrrrrrVrcSrrrrrrrrVrcSrr rrrrrrVrcSrrrrrrrrVrc s@g|]\}}dttggt|gt|fqS)rn)r^rurIrthgfoZformat_dense_tile)rRpositionZ tile_data) tileset_idrfr|rrrVs )rrr~rrtitchainrHrsrvrrrIfilterr/maxrmr)rrxrgZtileset_file_and_infoZtile_ids_by_zoom_and_transformZpartitioned_tile_idsrZ tile_groupZ tileset_fileZsorted_resolutionsr3reZtile_positionsZminxZmaxxZminyZmaxyZtile_data_by_positionrr)rrrfr|rrUs|           r)rN)rr r )& collectionsrpr*Zclodius.tiles.formatrformatrZclodius.tiles.utilsutilsrWrB itertoolsrnumpyr Zpandasr.logging getLogger__name__rFrrrJrrr8rPrXr`rmr~rvrrrrrrrrrsL  c6 C