o Dfk@s2dZddlZddlmZddlmZddlmZmZm Z m Z m Z m Z ddl mZmZddlmZeZGd d d eZGd d d eZd dZGdddZGdddZGdddeZGdddeeZGdddeZGdddeZdgfdefddZGdddeZ Gd d!d!e eZ!Gd"d#d#e!eZ"Gd$d%d%e"Z#Gd&d'd'e"Z$Gd(d)d)e"Z%Gd*d+d+e#Z&Gd,d-d-e$Z'Gd.d/d/e&Z(Gd0d1d1e'Z)Gd2d3d3eZ*Gd4d5d5e!Z+Gd6d7d7e Z,Gd8d9d9e eZ-Gd:d;d;e-Z.Gdd?Z0d@e edee1e1ffdAdBZ2dS)Cz Adapter finding and trimming classes The ...Adapter classes are responsible for finding adapters. The ...Match classes trim the reads. N)Enum) defaultdict)OptionalTupleSequenceDictAnyList)ABCabstractmethod)alignc@s eZdZdS)InvalidCharacterN)__name__ __module__ __qualname__rrZ/var/www/html/software/conda/envs/catlas/lib/python3.10/site-packages/cutadapt/adapters.pyrsrc@s^eZdZejejBejBZejejBejBZ ejZ ejZ ejejBZ ejejBZ ejZdZdS)WherelinkedN)rrrr ZSTART_WITHIN_SEQ2ZSTOP_WITHIN_SEQ2ZSTOP_WITHIN_SEQ1BACKZSTART_WITHIN_SEQ1FRONTPREFIXSUFFIXFRONT_NOT_INTERNALBACK_NOT_INTERNALZ SEMIGLOBALANYWHERELINKEDrrrrrs  rcCsttSN)rintrrrrreturns_defaultdict_int%sr c@sReZdZdZdddZddZdefd d Zed d Z d e de e fddZ dS) EndStatisticsz!Statistics about the 5' or 3' endadapter SingleAdaptercCsZ|j|_|j|_|j|_|j|_|j|_tt|_dddddd|_ t |t t t f|_dS)NrACGT)max_error_ratesequenceeffective_lengthadapter_wildcards has_wildcardsallows_partial_matchesrr errorsadjacent_bases isinstance FrontAdapterNonInternalFrontAdapter PrefixAdapter_remove_prefix)selfr"rrr__init__/s zEndStatistics.__init__cCs&dd|jD}d|j||jS)NcSsi|] \}}|t|qSr)dict).0kvrrr <sz*EndStatistics.__repr__..z>EndStatistics(max_error_rate={}, errors={}, adjacent_bases={}))r0itemsformatr*r1)r7r0rrr__repr__;s zEndStatistics.__repr__othercCst||js td|j|jks|j|jks|j|jkr tddD]}|j||j|7<q"|j D]\}}|D]}|j|||j||7<q)r7rAbaselengthZ error_dictr0rrr__iadd__Cs    "zEndStatistics.__iadd__cCsdd|jD}|S)NcSsi|] \}}|t|qSr)sumvalues)r:rFr0rrrr=Usz)EndStatistics.lengths..)r0r>)r7drrrlengthsSszEndStatistics.lengths gc_contentreturncCsx|j}|jr |ddd}|jrdnd}d}|g}t|D]\}}||vr,||d9}n|d|d9}||q|S)a Estimate probabilities that this adapter end matches a random sequence. Indels are not taken into account. Returns a list p, where p[i] is the probability that i bases of this adapter match a random sequence with GC content gc_content. NZ CGRYSKMBDHVNZGCg?g@)r+r6r. enumerateappend)r7rLseqZ allowed_basespZ probabilitiesicrrrrandom_match_probabilitiesXs  z(EndStatistics.random_match_probabilitiesN)r"r#) rrr__doc__r8r@rrGpropertyrKfloatr rUrrrrr!,s   r!c@s@eZdZdZ ddddddedfdd Zd d Zdd dZdS)AdapterStatisticsz Statistics about an adapter. An adapter can work on the 5' end (front) or 3' end (back) of a read, and statistics for that are captured separately in EndStatistics objects. Nr"Adapterfrontr#backcCs@|j|_||_t||_|durt||_nt||_d|_dSNr)namer"r!r[r\reverse_complemented)r7r"r[r\rrrr8xs    zAdapterStatistics.__init__cCd|j|j|jS)Nz-AdapterStatistics(name={}, front={}, back={}))r?r^r[r\r7rrrr@s zAdapterStatistics.__repr__rAcCs4|j|j7_|j|j7_|j|j7_|Sr)r[r\r_r7rArrrrGszAdapterStatistics.__iadd__r)rArY)rrrrVrr8r@rGrrrrrYqs  rYc@sleZdZUded<edeeeffddZedeeeffddZede e fdd Z ed d Z d S) MatchrZr"rMcCdSrrrarrrremainder_intervalzMatch.remainder_intervalcCrdrrrarrrretained_adapter_intervalrfzMatch.retained_adapter_intervalcCrdrrr7readrrrget_info_recordsrfzMatch.get_info_recordscCrdrrrhrrrtrimmedrfz Match.trimmedN) rrr__annotations__r rrrergr rjrkrrrrrcs rcc@s~eZdZdZgdZdedededededed d d efd d ZddZde fddZ ddedefddZ de e fddZ dS) SingleMatchzG Representation of a single adapter matched to a single string ) astartastoprstartrstopmatchesr0r"r+rF adjacent_basernrorprqrrr0r"r#r+c CsDd|_||_||_||_||_||_||_||_||_|||_ dS)Nr)) rsrnrorprqrrr0r"r+rF) r7rnrorprqrrr0r"r+rrrr8s zSingleMatch.__init__cC d|j|j|j|j|j|jS)NzLSingleMatch(astart={}, astop={}, rstart={}, rstop={}, matches={}, errors={})r?rnrorprqrrr0rarrrr@zSingleMatch.__repr__rMcCsl|j|juo5|j|jko5|j|jko5|j|jko5|j|jko5|j|jko5|j|jko5|j|juo5|j|jkSr) rBrnrorprqrrr0r"r+rbrrr__eq__s"         zSingleMatch.__eq__N wildcard_charcs$fddtjD}d|S)a4 Return a string that contains, for each wildcard character, the character that it matches. For example, if the adapter ATNGNA matches ATCGTA, then the string 'CT' is returned. If there are indels, this is not reliable as the full alignment is not available. csFg|]}jjj|kr!j|tjkrjj|qSr)r"r+rnrplen)r:rSr7ryrr s z)SingleMatch.wildcards..r))rangerFjoin)r7ry wildcardsrr{rrs zSingleMatch.wildcardsc Cs|j}|j}d|j|j|j|d|j||j|j||jd|jjg}|rA||d|j||j|j||jdg7}|gS|gd7}|gS)Nr)r)r)r)r))r+ qualitiesr0rprqr"r^)r7rirQrinforrrrjs(      zSingleMatch.get_info_recordsN)rx)rrrrV __slots__rstrr8r@boolrwrr rjrrrrrms0  rmc@peZdZdZddZdefddZdeeeffddZ deeeffd d Z d d Z d dZ de fddZdS)RemoveBeforeMatchz.A match that removes sequence before the matchcCrt)NzRRemoveBeforeMatch(astart={}, astop={}, rstart={}, rstop={}, matches={}, errors={})rurarrrr@rvzRemoveBeforeMatch.__repr__rMcCs|jd|jSz Return the part of the read before this match if this is a 'front' (5') adapter, return the part after the match if this is not a 'front' adapter (3'). This can be an empty string. N)r+rprarrrrest zRemoveBeforeMatch.restcC|jt|jfS) Return an interval (start, stop) that describes the part of the read that would remain after trimming )rqrzr+rarrrresz$RemoveBeforeMatch.remainder_intervalcCrr)rprzr+rarrrrgz+RemoveBeforeMatch.retained_adapter_intervalcCs t|jdSr)slicerqrarrr trim_slice zRemoveBeforeMatch.trim_slicecCs||jdSrrqrhrrrrk zRemoveBeforeMatch.trimmed statisticscCs |jj|j|jd7<dS!Update AdapterStatistics in placer N)r[r0rq)r7rrrrupdate_statistics#s z#RemoveBeforeMatch.update_statisticsNrrrrVr@rrrrrergrrkrYrrrrrr rc@r)RemoveAfterMatchz-A match that removes sequence after the matchcCrt)NzQRemoveAfterMatch(astart={}, astop={}, rstart={}, rstop={}, matches={}, errors={})rurarrrr@+rvzRemoveAfterMatch.__repr__rMcCs|j|jdSr)r+rqrarrrr/rzRemoveAfterMatch.restcC d|jfS)rrrprarrrre8s z#RemoveAfterMatch.remainder_intervalcCrr]rrarrrrg? z*RemoveAfterMatch.retained_adapter_intervalcCs td|jSr)rrprarrrrBrzRemoveAfterMatch.trim_slicecCs|d|jSrrrhrrrrkFrzRemoveAfterMatch.trimmedrcCsx|j|jd|j}|jjt|j|j|jd7<z |jj|d7<WdSty;d|jjd<YdSw)rr r)N)r+rpr\r0rzr1KeyError)r7rrsrrrrIs& z"RemoveAfterMatch.update_statisticsNrrrrrr(rrrMcCs t|d}|dd7<|S)Nrr )r)_startr^rrr_generate_adapter_nameSs rc@s<eZdZdZdefddZeddZedefdd Zd S) Matchablez'Something that has a match_to() method.r^cOs ||_dSrr^)r7r^argskwargsrrrr8\rzMatchable.__init__cCrdrrrarrr enable_debug_rfzMatchable.enable_debugr+cCrdrrr7r+rrrmatch_tocrfzMatchable.match_toN) rrrrVrr8r rrrrrrrYs rc@s"eZdZdZedefddZdS)rZzadapter with one componentrMcCrdrrrarrrcreate_statisticslrfzAdapter.create_statisticsN)rrr descriptionr rYrrrrrrZhsrZcseZdZUdZdZeed<      d#ded ed e d ed ed e edeffdd Z de de j fddZddZede fddZd$ddZeddZedefddZde fdd Zdefd!d"ZZS)%r#aw This class can find a single adapter characterized by sequence, error rate, type etc. within reads. where -- A Where enum value. This influences where the adapter is allowed to appear within the read. sequence -- The adapter sequence as string. Will be converted to uppercase. Also, Us will be converted to Ts. max_errors -- Maximum allowed errors (non-negative float). If the values is less than 1, this is interpreted as a rate directly and passed to the aligner. If it is 1 or greater, the value is converted to a rate by dividing it by the length of the sequence. The error rate is the number of errors in the alignment divided by the length of the part of the alignment that matches the adapter. minimum_overlap -- Minimum length of the part of the alignment that matches the adapter. read_wildcards -- Whether IUPAC wildcards in the read are allowed. adapter_wildcards -- Whether IUPAC wildcards in the adapter are allowed. name -- optional name of the adapter. If not provided, the name is set to a unique number. Tr/皙?FNr+ max_errors min_overlapread_wildcardsr-r^indelsc s|durtn||_t|jd|_|dd|_|js$td|dkr/|t |j}||_ t |t |j|_ t d}|r[t|j|ks[|jD]} | |vrZtd| |jqK|oft|jtdk |_||_||_||_dS) NFUr(zAdapter sequence is emptyr ZABCDGHKMNRSTUVWXYziCharacter {!r} in adapter sequence {!r} is not a valid IUPAC code. Use only characters ABCDGHKMNRSTUVWXY.ZACGT)rr^superr8_debugupperreplacer+rCrzr*minr frozensetsetrr?r-rr_aligneraligner) r7r+rrrr-r^rZiupacrTrBrrr8s,   zSingleAdapter.__init__flagsrMc Cs0|jrdnd}tj|j|j||j|j||jdS)Nr i)r wildcard_refwildcard_query indel_costr)rr Alignerr+r*r-rr)r7rrrrr _make_alignerszSingleAdapter._make_alignercCsdjdd|jjit|S)Nz<{cls}(name={name!r}, sequence={sequence!r}, max_error_rate={max_error_rate}, min_overlap={min_overlap}, read_wildcards={read_wildcards}, adapter_wildcards={adapter_wildcards}, indels={indels})>clsr)r?rBrvarsrarrrr@szSingleAdapter.__repr__cC|jjSr)rr,rarrrr,zSingleAdapter.effective_lengthcCsd|_|jdS)zg Print out the dynamic programming matrix after matching a read to an adapter. TN)rrrrarrrrszSingleAdapter.enable_debugcCrdrrrarrrrrfzSingleAdapter._alignercCdS) Attempt to match this adapter to the given string. Return a Match instance if a match was found; return None if no match was found given the matching criteria (minimum overlap length, maximum error rate). NrrrrrrzSingleAdapter.match_tocC t|jSr)rzr+rarrr__len__rzSingleAdapter.__len__cCs t||Sr)rYrarrrrrzSingleAdapter.create_statistics)rrFTNT)rMN)rrrrVr/rrlrrXrrr8r rrr@rWr,rr rrrrYr __classcell__rrrrr#qsF  !   r#csBeZdZdZdZfddZdejfddZde fd d Z Z S) r3z A 5' adapterz regular 5'c$|dd|_tj|i|dSNZforce_anywhereFpop_force_anywhererr8r7rrrrrr8zFrontAdapter.__init__rMcC||jr tjjStjjSr)rrrrvaluerrarrrrzFrontAdapter._alignerr+cC:|j|}|jrt|jj|durdSt|||dSz Attempt to match this adapter to the given read. Return a Match instance if a match was found; return None if no match was found given the matching criteria (minimum overlap length, maximum error rate). Nr"r+)rlocaterprintdpmatrixrr7r+ alignmentrrrr  zFrontAdapter.match_to) rrrrVrr8r rrrrrrrrrr3s  r3cs:eZdZdZdZfddZddZdefdd ZZ S) BackAdapterz A 3' adapterz regular 3'crrrrrrrr8 rzBackAdapter.__init__cCrr)rrrrrrrarrrrrzBackAdapter._alignerr+cCrr)rrrrrrrrrrrrzBackAdapter.match_to) rrrrVrr8rrrrrrrrrs  rc@*eZdZdZdZddZdefddZdS) AnywhereAdapterz An adapter that can be 5' or 3'. If a match involves the first base of the read, it is assumed to be a 5' adapter and a 3' otherwise. zvariable 5'/3'cC|tjjSr)rrrrrarrrr)rzAnywhereAdapter._alignerr+cCsb|j|}|jrt|jj|durdS|ddkr't|||d}|St|||d}|S)rNrr)rrrrrrrr)r7r+rmatchrrrr,s  zAnywhereAdapter.match_toNrrrrVrrrrrrrrr!s rc@r) r4zA non-internal 5' adapterznon-internal 5'cCrr)rrrrrarrrrFrz NonInternalFrontAdapter._alignerr+cCR|j|}|jrzt|jjWn tyYnw|dur!dSt|||dSNr)rrrrrAttributeErrorrrrrrrI  z NonInternalFrontAdapter.match_toNrrrrrr4A r4c@r) NonInternalBackAdapterzA non-internal 3' adapterznon-internal 3'cCrr)rrrrrarrrr[rzNonInternalBackAdapter._alignerr+cCrr)rrrrrrrrrrrr^rzNonInternalBackAdapter.match_toNrrrrrrVrrc@ eZdZdZdZdZddZdS)r5zAn anchored 5' adapterz anchored 5'FcC2|jstj|j|j|j|j|jdS|t j j SN)rrr) rr ZPrefixComparerr+r*r-rrrrrrrarrrrqzPrefixAdapter._alignerNrrrrVrr/rrrrrr5k  r5c@r) SuffixAdapterzAn anchored 3' adapterz anchored 3'FcCrr) rr ZSuffixComparerr+r*r-rrrrrrrarrrrrzSuffixAdapter._alignerNrrrrrr~rrc@seZdZdZdededdfddZdd Zed d Z ed d Z ddZ eddZ ddZ deeeffddZdeeeffddZdeefddZdS) LinkedMatchz. Represent a match of a LinkedAdapter front_match back_matchr" LinkedAdaptercCs*|dus |dus J||_||_||_dSr)rrr")r7rrr"rrrr8s zLinkedMatch.__init__cCr`)Nz:)r?rrr"rarrrr@s zLinkedMatch.__repr__cCs4d}|jdur ||jj7}|jdur||jj7}|S)zNumber of matching basesrN)rrrr)r7mrrrrrs     zLinkedMatch.matchescCs4d}|jdur ||jj7}|jdur||jj7}|Sr])rr0r)r7errrr0s     zLinkedMatch.errorscCs(|jr |j|}|jr|j|}|Sr)rrkrrhrrrrks   zLinkedMatch.trimmedcCrr)rrsrarrrrsrzLinkedMatch.adjacent_basecCsd|jr|jj|jj|jjd7<|jr0t|jj|jj}|jj||jjd7<dSdSr) rr[r0rqrrzr+rpr\)r7rrFrrrrs   zLinkedMatch.update_statisticsrMcCsdd|j|jfD}t|S)NcSsg|]}|dur|qSrr)r:rrrrr|sz2LinkedMatch.remainder_interval..)rr remainder)r7rrrrrreszLinkedMatch.remainder_intervalcCsN|jr |jj}|jj}nd}}|jr|jj|}||fSt|jj}||fSr])rrprqrrzr+)r7startoffsetendrrrrgs   z%LinkedMatch.retained_adapter_intervalcCsrg}|jdf|jdffD]*\}}|durq ||d}|jjdur$dn|jj||d<||||}q |S)Nz;1z;2rnone)rrrjr"r^rPrk)r7rirecordsrZ namesuffixrecordrrrrjs    zLinkedMatch.get_info_recordsN)rrrrVrrr8r@rWrrr0rkrsrrrrergr rjrrrrrs    rc s~eZdZdZdZdededededef fdd Zd d Z d ed e e fddZ d e fddZeddZeddZZS)rz'A 5' adapter combined with a 3' adapterr front_adapter back_adapterfront_required back_requiredr^csNt|||_||_tj|_|durtn||_||_ |j|j _||_ dSr) rr8rrrrwhererr^rr)r7rrrrr^rrrr8s   zLinkedAdapter.__init__cCs|j|jdSr)rrrrarrrrs zLinkedAdapter.enable_debugr+rMcCsd|j|}|jr|durdS|dur||}|j|}|dur,|js*|dur,dSt|||S)z@ Match the two linked adapters against a string N)rrrrrrr)r7r+rrrrrrs    zLinkedAdapter.match_tocCst||j|jSr)rYrrrarrrrrzLinkedAdapter.create_statisticscCs|jjd|jjS)Nz...)rr+rrarrrr+szLinkedAdapter.sequencecCrdrrrarrrremoverfzLinkedAdapter.remove)rrrrVrr#rrr8rrrrrYrrWr+rrrrrrrs* rcsXeZdZdZdeeffdd ZddZddZd d Z d e d e e fd dZ ZS)MultipleAdaptersz- Represent multiple adapters at once adapterscstjdd||_dS)NZmultiple_adaptersr)rr8 _adapters)r7rrrrr8s zMultipleAdapters.__init__cCs|jD]}|qdSr)rr)r7arrrr"s  zMultipleAdapters.enable_debugcCs |j|Sr)r)r7itemrrr __getitem__&rzMultipleAdapters.__getitem__cCrr)rzrrarrrr)rzMultipleAdapters.__len__r+rMcCsXd}|jD]$}||}|durq|dus'|j|jks'|j|jkr)|j|jkr)|}q|S)z Find the adapter that best matches the sequence. Return either a Match instance or None if there are no matches. N)rrrrr0)r7r+ best_matchr"rrrrr,s  zMultipleAdapters.match_to)rrrrVrrr8rr rrrrmrrrrrrrsrcseZdZdZeeeeeeffZ fddZ ddZ defddZ e d d Ze d efd d ZeddZeddZd eeedffddZdefddZdefddZddZZS)IndexedAdaptersa Represent multiple adapters of the same type at once and use an index data structure to speed up matching. This acts like a "normal" Adapter as it provides a match_to method, but is faster with lots of adapters. There are quite a few restrictions: - the error rate allows at most 2 mismatches - wildcards in the adapter are not allowed - wildcards in the read are not allowed Use the is_acceptable() method to check individual adapters. cstjdd|s td|D]}||q||_t||_|\|_|_ t dt |jddt |jdkrD|jd|_|j|_n|j|_||_d S) z+All given adapters must be of the same typeZindexed_adaptersrzAdapter list is emptyzString lengths in the index: %sTreverser rN)rr8rC_acceptrr_multiple_adapters _make_index_lengths_indexloggerdebugsortedrz_length_match_to_one_lengthr_match_to_multiple_lengths_get_make_affix _make_affix)r7rr"rrrr8Os    zIndexedAdapters.__init__cCsd|jj|jS)Nz{}(adapters={!r}))r?rBrrrarrrr@aszIndexedAdapters.__repr__r+cCr)z4Never called because it gets overwritten in __init__NrrrrrrdrzIndexedAdapters.match_tocCrdrrrarrrrgrfzIndexedAdapters._get_make_affixrMcCrdrrr7r"rFrrr0r+rrr _make_matchkrfzIndexedAdapters._make_matchcCsB|jrtd|jrtdtt||j}|dkrtddS)z3Raise a ValueError if the adapter is not acceptablez#Wildcards in the read not supportedz&Wildcards in the adapter not supportedrzError rate too highN)rrCr-rrzr*)rr"r;rrrroszIndexedAdapters._acceptcCs&z||WdStyYdSw)z Return whether this adapter is acceptable for being used in an index Adapters are not acceptable if they allow wildcards, allow too many errors, or would lead to a very large index. FT)rrCrr"rrr is_acceptablezs   zIndexedAdapters.is_acceptable AdapterIndexc Cstdt|jt}t}d}|jD]Y}|j}t|jt|}|j r(t j nt j }|||D]<\}} } ||vr^||\} } } | | krEq0| | kr]|s]t d| j| j|j|j||| d}n|| | f||<|t|q0qtdt|t|dd|fS)Nz!Building index of %s adapters ...FzAdapters %s %r and %s %r are very similar. At %s allowed errors, the sequence %r cannot be assigned uniquely because the number of matches is %s compared to both adapters.Tz%Built an index containing %s strings.r )rrrzrr9rr+rr*rr Zedit_environmentZhamming_environmentwarningr^addr)r7indexrKZ has_warnedr"r+r; environmentsr0rrZ other_adapterZ other_errorsZ other_matchesrrrrs4    zIndexedAdapters._make_indexcCsd|||j}d|vr|j|Sz |j|\}}}Wn ty'YdSw|||j|||S) Match the adapters against a string and return a Match that represents the best match or None if no match was found rxN)rrrrrrrr)r7r+affixr"rrrrrrs  z$IndexedAdapters._match_to_one_lengthc Cs|}d}d}d}d}d}|jD]F}||krn?|||}|r/d|vr-|j|Sd}z |j|\} } } Wn tyBYqw| |ksO| |krW| |krW| }| }| }|}q|dkr^dS||||||S)r%NrrNiTrxF)rrrrrrrr) r7r+r&Z best_adapterZ best_lengthZbest_mZbest_eZcheck_nrFr"rrrrrrs8   z*IndexedAdapters._match_to_multiple_lengthscCrdrrrarrrrszIndexedAdapters.enable_debug)rrrrVrrrr#rrr8r@rr rrmr classmethodrrr rrrrrrrrrr @s$     'r c<eZdZefddZddZddZeddZZ S) IndexedPrefixAdaptersct|ts tdt|S)Nz%Only 5' anchored adapters are allowed)r2r5rCrrrrrrr  zIndexedPrefixAdapters._acceptc Cstdt|jd|||||dSNr)rnrorprqrrr0r"r+)rrzr+rrrrrsz!IndexedPrefixAdapters._make_matchcC|jSr) _make_prefixrarrrrz%IndexedPrefixAdapters._get_make_affixcCs |d|Srrr$nrrrr.rz"IndexedPrefixAdapters._make_prefix) rrrr'rrr staticmethodr.rrrrrr) r)cr() IndexedSuffixAdapterscr*)Nz%Only anchored 3' adapters are allowed)r2rrCrrrrrrrr+zIndexedSuffixAdapters._acceptc Cs*tdt|jt||t|||||dSr,)rrzr+rrrrrs z!IndexedSuffixAdapters._make_matchcCr-r) _make_suffixrarrrrr/z%IndexedSuffixAdapters._get_make_affixcCs|| dSrrr0rrrr5sz"IndexedSuffixAdapters._make_suffix) rrrr'rrrr2r5rrrrrr4r3r4cCsDt}|D]}|j|jf}||vrtd|j|j|j||<qdS)NzZAdapter %r (%s) was specified multiple times! Please make sure that this is what you want.)r9rBr+rr rr^)rrJr"keyrrrwarn_duplicate_adapterss  r7rrcCsB|stdd}|D] }|\}}||7}q ||}|||fS)z Determine which section of the read would not be trimmed. Return a tuple (start, stop) that gives the interval of the untrimmed part relative to the original read. matches must be non-empty zmatches must not be emptyr)rCre)rrrrZ match_startZ match_stoprFrrrr#s   r)3rVloggingenumr collectionsrtypingrrrrrr abcr r r)r getLoggerr Exceptionrrr r!rYrcrmrrrrrrZr#r3rrr4rr5rrrrr r)r4r7rrrrrrsJ    E$X&+ x S6&!"