ó ¤>dc@s­ddlmZddlZddlZddlmZddlmZm Z m Z ddl m Z m Z mZddlmZdd d„ƒYZd efd „ƒYZdS( iÿÿÿÿ(t ConfigParserN(tdeque(tVBZtraise_missing_vbz_error_readtraise_missing_vbz_error_write(t_sanitize_data_for_readingt_sanitize_data_for_writingt_clean(tLEGACY_COMPONENT_NAMESt AbstractFast5cBs_eZd„Zd„Zd„Zd„Zd„Zed„ƒZd„Z d„Z d„Z RS( cCs|S(N((tself((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyt __enter__ scCs|jƒtS(N(tclosetFalse(R texception_typetexception_valuet traceback((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyt__exit__s cCs tƒ‚dS(N(tNotImplementedError(R ((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyt get_readsscCs tƒ‚dS(N(R(R ((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyt get_read_idsscCs tƒ‚dS(N(R(R tread_id((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pytget_readscCst|jƒdkS(Nu(treprthandle(R ((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyt_is_openscCs(|js$tdj|jƒƒ‚ndS(NsFast5 file is not open: {}(RtIOErrortformattfilename(R ((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyt assert_open!s cCs>|jƒ|jdkr:tdj|j|jƒƒ‚ndS(Ntrs'Fast5 file is in read-only mode '{}' {}(RtmodeRRR(R ((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pytassert_writeable%s cCsM|jrId|_|jr4|jjƒd|_nd|_d|_ndS(s Closes the object. N(RtNoneRRR Rtstatus(R ((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyR *s      ( t__name__t __module__R RRRRtpropertyRRR R (((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyR s       t Fast5ReadcBsßeZdZd„Zd„Zd„Zed„ƒZed„ƒZed„ƒZ ed„ƒZ d*e d„Z ed „Zed „Zd*d „Zd*d*d*ed „Zd*d „Zed„Zd„Zd„Zd„Zed„Zd„Zed„ƒZd„Zed„Zd„Zd„Zd„Zd„Z d„Z!d„Z"d*d„Z#ed„Z$d„Z%d*d „Z&ed!„Z'e(d"„ƒZ)e(d#„ƒZ*d$„Z+d%„Z,ed&„Z-d'„Z.d(„Z/d)„Z0RS(+tcCsB||_|j|_|j|_|jd||_||_dS(Ntread_(tparentRRRR(R R)R((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyt__init__9s    cCstdƒ|jS(NsN'read.get_run_id()' will be deprecated. Use the property 'read.run_id' instead(tDeprecationWarningtrun_id(R ((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyt get_run_idAs cCstdƒ|jS(NsP'read.get_read_id()' will be deprecated. Use the property 'read.read_id' instead(R+R(R ((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyt get_read_idEs cCs |jdS(Ns/Signal(traw_dataset_group_name(R ((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pytraw_dataset_nameIscCsdS(NtRaw((R ((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyR/MscCs|j|jjS(N(RR0t_filters(R ((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pytraw_compression_filtersQscCs|j|jdjdS(Nt tracking_idR,(Rt global_keytattrs(R ((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyR,UscCsç|jƒ|j|jkr2|jj|jƒn|j|jkrkd}t|j|j|jƒƒ‚ny0|j|jj dd|ddt |ƒWnt k rº}t |ƒnX|dk rã|j|j|dtƒndS(sÀ Add raw data for a read. :param data: The raw data DAQ values (16 bit integers). The read must already exist in the file. It must not already have raw data. s4Fast5 file already has raw data for read '{}' in: {}tSignaltdatatdtypeti2tclearN(R R/Rt create_groupR0tKeyErrorRRRtcreate_datasettvarst ValueErrorRR!t_add_attributestTrue(R R8R6t compressiontmsgte((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyt add_raw_dataYs ! cCsC|jƒd|jkr,|jjdƒn|jd||ƒdS(s Add channel info data to the channel_id group. :param data: A dictionary of key/value pairs. Keys must be strings. Values can be strings or numeric values. :param clear: If set, any existing channel info data will be removed. t channel_idN(R RR<RA(R R6R;((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pytadd_channel_infops cCs@|jƒd|jkr,|jjdƒn|j||ƒdS(NR4(R RR<tset_tracking_id(R R6R;((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pytadd_tracking_id|s cCs»|jƒd|jkr,|jjdƒndj|ƒ}dj|ƒ}|jƒ}||d<|j||ƒ|j|jdƒ|j|jdƒ|dk r·|j||ƒndS(sh Add a new analysis group to the file. :param component: The component name. :param group_name: The name to use for the group. Must not already exist in the file e.g. 'Test_000'. :param attrs: A dictionary containing the key-value pairs to put in the analysis group as attributes. Keys must be strings, and values must be strings or numeric types. :param config: A dictionary of dictionaries. The top level keys should be the name of analysis steps, and should contain key value pairs for analysis parameters used. tAnalysess Analyses/{}s{}/Configurationt componenttSummaryt ConfigurationN(R RR<Rtcopyt _add_groupR!t_add_attribute_tree(R RLt group_nameR6tconfigtgroupt cfg_groupt group_attrs((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyt add_analysis‚s    cCs/|rtjdƒn|j|j|||ƒS(Ns*Read number is not used for MultiReadFast5(twarningstwarnt _load_rawR0(R t read_numbertstarttendtscale((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyt get_raw_dataœscCsä|jƒg}d|jkr#|S|jdjƒ}x§|D]Ÿ}|jdj|ƒj}d|kr{t|dƒ}n'|d tkrœt|d }nd}|dk r=|dksÆ||kr=|j||fƒq=q=W|S(sö Provides a list of all analyses groups. :param component: Optional component name. If provided, only analyses of that component will be returned. :returns: A list of component-name/group-name pairs (tuples). RKs Analyses/{}RLiüÿÿÿN( RRtkeysRR6RRR!tappend(R RLtanalysest ana_groupsRRRVtcomp((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyt list_analyses¡s   $cCsÛ|jƒ|jƒ}g}x6|D].}|dd |kr#|j|dƒq#q#Wt|ƒdkr‰d }|r…dj|ƒ}n|St|ƒd}|r×t|dƒd}dj|t|ƒj dƒƒ}n|S( sL Return the latest analysis group with the specified base name. :param group_base: The name of the analysis group, excluding the _### at the end e.g. 'Alignment' rather than 'Alignment_000' :param increment: If set, this will find the name of the latest analysis group, increment the index by 1, and return the results. If there aren't any, it will give an index of 000. :returns: The group name with the highest index, or None if there are no analyses with the specified base name. iiüÿÿÿis{}_000iÿÿÿÿiýÿÿÿs{}_{}iN( RReRatlenR!Rtsortedtinttstrtzfill(R t group_baset incrementt all_analysestselectedtanalysistresulttcount((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pytget_latest_analysisºs   $cCsf|jƒxE|jƒD]7\}}|jdƒsDdj|ƒ}n||| Returns a dictionary of tracking-id key/value pairs. R4cSs%i|]\}}t|ƒ|“qS((R(t.0R€R((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pys s (RRR5R6Rs(R ttracking((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pytget_tracking_ids cCs(|jƒ|j|jd||ƒdS(s Add tracking-id data to the tracking_id group. :param data: A dictionary of key/value pairs. Keys must be strings. Values can be strings or numeric values. :param clear: If set, any existing tracking-id data will be removed. R4N(R RAR5(R R8R;((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyRIs cCsO|jƒ|j|jdjjƒ}d„|Dƒ}t|dƒ|d<|S(sF Returns a dictionary of channel information key/value pairs. RGcSs%i|]\}}t|ƒ|“qS((R(R„R€R((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pys ,s tchannel_number(RRR5R6RsRh(R t channel_info((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pytget_channel_info's  cCs d|jkS(Nt context_tags(R(R ((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pythas_context_tags0scCsB|jƒ|jr>|j|jdjjƒ}d„|DƒSiS(s> Returns a dictionary of context tag key/value pairs. RŠcSs%i|]\}}t|ƒ|“qS((R(R„R€R((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pys :s (RR‹RR5R6Rs(R ttags((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pytget_context_tags4s   cCsK|jƒ|jr0|j|jd||ƒn|j|jd|ƒdS(s Replaces any existing context tag data with the provided values. :param data: A dictionary of key/value pairs. Keys must be strings. Values can be strings or numeric values. :param clear: If set, any existing context tag data will be removed. RŠN(R R‹RAR5RP(R R8R;((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pytadd_context_tags=s  cCstdƒ‚dS(Ns[Cannot add_existing_read() to a Fast5Read(). Use MultiFast5File.create_empty_read() instead(R(R R[Rt start_timetdurationtmuxt median_before((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pytadd_readJscCsV|jƒ||jkr,|j|iƒnt|ƒ}|j|j|d|ƒdS(sí Add a log the file. :param group_name: Global group name. Cave: No 'Analyses' is prepended. :param field_name: A field {group_name}/{field_name} will be written. :param log_string: content to be written. R8N(R RRPRR>(R RRt field_namet log_stringtsanitized_data((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pytadd_logNs   cCs0|jƒdj||ƒ}|j||ƒdS(sg Set the summary data for an analysis group. :param group_name: The name of the analysis group. :param section_name: The analysis step. This will become a subfolder in the Summary section. :param data: A dictionary containing keys which are the summary fields, and values which are the summary values. sAnalyses/{}/Summary/{}N(R RRP(R RRt section_nameR8RT((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pytset_summary_dataZs cCsè|jƒdj|ƒ|jkrFd}t|j||jƒƒ‚nt|tƒr’i}xg|jƒD]#}d„|j|ƒDƒ||zs s,config must be a ConfigParser or dict not {}sAnalyses/{}/ConfigurationN( R RRR=Rt isinstanceRtsectionsRstdictt TypeErrorttypeRQ(R RRRSRDt config_dicttsectiont config_path((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pytset_analysis_configgs $ cCsD|jƒdj|ƒ}d}||jkr@|j|ƒ}n|S(sx Gets any config data saved for the analysis. :param group_name: The name of the analysis group. :returns: A dictionary of dictionaries. Each key represents an analysis step. Each value is a dictionary containing the analysis parameters as key/value pairs. Returns None if no configuration exists for the analysis. sAnalyses/{}/ConfigurationN(RRR!Rt_parse_attribute_tree(R RRRTRS((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pytget_analysis_configƒs cCsD|jƒdj|ƒ}d}||jkr@|j|ƒ}n|S(s? Get the summary data for an analysis group. :param group_name: The name of the analysis group to pull summary data for. :returns: A dictionary whose keys are analysis steps, and whose values are dictionaries of key/value pairs for the results of that step. sAnalyses/{}/SummaryN(RRR!RR¥(R RRRTtsummary((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pytget_summary_data“s cCs0|jƒdj||ƒ}|j||ƒdS(s Add a new subgroup to an existing analysis group. :param group_name: The name of the analysis group you are adding to. :param subgroup_name: The name of the new subgroup. :param attrs: A dictionary representing the attributes to assign to the subgroup. The new subgroup must not already exist. The subgroup name can be a nested name, such as "Template/Data". This will create the "Template" subgroup (if it does not exist), and the "Data" subgroup below it. sAnalyses/{}/{}N(R RRP(R RRt subgroup_nameR6RT((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pytadd_analysis_subgroup£s cCs0|jƒdj|ƒ}|j|||ƒdS(s Add attributes on the group or dataset specified. :param group_name: The name of the group (or dataset). :param attrs: A dictionary representing the attributes to add. :param clear: If set, any existing attributes will be cleared. The specified group name can be any existing path (relative to the "Analyses" group. It can be a group or a dataset. s Analyses/{}N(R RRA(R RRR6R;RT((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyRuµs cCs[|jƒdj|ƒ}d}||jkrW|j|jjƒ}d„|Dƒ}n|S(sõ Returns the attributes for the specified group or dataset. :param group_name: The path of the group or dataset, relative to the "Analyses" group. :returns: A dictionary representing the attributes (if any). s Analyses/{}cSs%i|]\}}t|ƒ|“qS((R(R„R€R((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pys Ïs N(RRR!RR6Rs(R RRRTR}((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pytget_analysis_attributesÃs c Csß|jƒdj|ƒ}||jkrLd}t|j||jƒƒ‚nt|ƒ}tj|ƒdkrŠ|j|j|d|ƒn |j|j|d|ddƒ|dk rÛdj||ƒ}|j ||ƒndS( sö Add a dataset to the specified group. :param group_name: The path of the group the dataset will be added to, relative to the "Analyses" group. :param dataset_name: The name of the new dataset. :param data: A numpy array representing the data to be written. :param attrs: A dictionary of attributes to be added to the dataset. :raises KeyError: If dataset is being added to non-existant group or if file is not open for writing. s Analyses/{}s@Dataset cannot be added to non-existent group: Analyses/{} in {}R8RCtgzips{}/{}N(( R RRR=RRtnptshapeR>R!RA( R RRt dataset_nameR8R6t group_pathRDR–Rw((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pytadd_analysis_datasetÒs    cCsZ|jƒdj||ƒ}d}||jkrV|j|}|sVt|ƒ}qVn|S(s Return the specified dataset, as a numpy array. :param group_name: The path of the group containing the dataset, relative to the "Analyses" group. :param dataset_name: The name of the dataset. :param skip_decoding: If True, this will directly return the h5py dataset. If False, it will load the data into a Numpy array and decode any byte-strings :return: A numpy array containing the data. Returns None if the dataset does not exist. sAnalyses/{}/{}N(RRR!RR(R RRR¯t skip_decodingR8((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pytget_analysis_datasetïs  cCstdƒ‚dS(Ns;read_summary_data() is not implemented for MultiFast5 reads(R(tfnameRL((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pytread_summary_datascCstdƒ‚dS(Ns.Cannot update legacy file for a MultiFast5Read(R(R´((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pytupdate_legacy_filesc Cs¤y|j|||!}Wntk r7}t|ƒnX|r |j|jdj}|d}|d} |d} | |} tj| || dtjƒ}n|S(NRGt digitisationtrangetoffsetR9(RtOSErrorRR5R6R­tarraytfloat32( R R¯R\R]R^trawterrRˆtdigitparangeR¹tscaling((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyRZ s    #cCs3|jj|ƒ|dk r/|j||ƒndS(sQ :param group: group_name :param attrs: :return: N(RR<R!RA(R RTR6((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyRPs cCsh|j|}|j}|r=x|jƒD] }||=q)Wnx$|jƒD]\}}|||Bs (RR`RR6Rs(R RTR8tfoldersRÅRwR}((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyR¥<s N(1R#R$R5R*R-R.R%R0R/R3R,R!RRFR RHRJRWR_ReRrRxRƒR†RIR‰R‹RRŽR“R—R™R¤R¦R¨RªRuR«R±R³t staticmethodRµR¶RZRPRARÄRQR¥(((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyR&6sT         !             ((t configparserRtnumpyR­RXt collectionsRt"ont_fast5_api.compression_settingsRRRtont_fast5_api.data_sanitisationRRRtont_fast5_api.static_dataRR R&(((sA/tmp/pip-install-m8tpmx/ont-fast5-api/ont_fast5_api/fast5_read.pyts  *