Word entropy-based approach to detect highly variable genetic markers for bacterial genotyping

ORIGINAL RESEARCH published: 03 Feb ua y 2021 doi: 10.3389/ micb.2021.631605 F on ie s in Mic obiology | www. on ie sin.o g 1Feb ua y 2021 | Volume 12 | A icle 631605 Edi ed by: Dominique La enie , UMR6074 Ins i u de Reche che en In o ma ique e Sys èmes Aléa oi es (IRISA), F ance Re iewed by: Mahend a Ma iadassou, Ins i u Na ional de Reche che pou l’Ag icul u e, l’Alimen a ion e l’En i onnemen (INRAE), F ance William She win, Uni e si y o New Sou h Wales, Aus alia *Co espondence: Ma ke a Nyk yno a nyk yno a@ u b .cz Special y sec ion: This a icle was submi ed o E olu iona y and Genomic Mic obiology, a sec ion o he jou nal F on ie s in Mic obiology Recei ed: 20 No embe 2020 Accep ed: 13 Janua y 2021 Published: 03 Feb ua y 2021 Ci a ion: Nyk yno a M, Ba on V, Sedla K, Bezdicek M, Lenge o a M and Sku ko a H (2021) Wo d En opy-Based App oach o De ec Highly Va iable Gene ic Ma ke s o Bac e ial Geno yping. F on . Mic obiol. 12:631605. doi: 10.3389/ micb.2021.631605 Wo d En opy-Based App oach o De ec Highly Va iable Gene ic Ma ke s o Bac e ial Geno yping Ma ke a Nyk yno a1*, Voj ech Ba on1, Ka el Sedla 1, Ma ej Bezdicek2, Ma ina Lenge o a2and Helena Sku ko a1 1Depa men o Biomedical Enginee ing, Facul y o Elec ical Enginee ing and Communica ion, B no Uni e si y o Technology, B no, Czechia, 2Depa men o In e nal Medicine - Hema ology and Oncology, Uni e si y Hospi al B no, B no, Czechia Geno yping me hods a e used o dis inguish bac e ial s ains om one species. Thus, dis inguishing bac e ial s ains on a global scale, be ween coun ies o local dis ic s in one coun y is possible. Howe e , he highly selec ed bac e ial popula ions (e.g., local popula ions in hospi al) a e ypically closely ela ed and low di e si ied. The e o e, cu en ly used yping me hods a e no able o dis inguish indi idual s ains om each o he . He e, we p esen a no el pipeline o de ec highly a iable gene ic segmen s o geno yping a closely ela ed bac e ial popula ion. The me hod is based on a deg ee o diso de in analyzed sequences ha can be ep esen ed by sequence en opy. Wi h he iden i ied a iable sequences, i is possible o ind ou ansmission ou es and sou ces o highly i ulen and mul i esis an s ains. The p oposed me hod can be used o any bac e ial popula ion, and due o i s whole genome ange, also non-coding egions a e examined. Keywo ds: geno yping, en opy, gene ic ma ke s, closely ela ed bac e ia, MLST 1. INTRODUCTION Heal hca e-associa ed in ec ions (HAIs) a e a se ious wo ldwide h ea wi h signi ican impac on mo ali y and mo bidi y o hospi alized pa ien s (A e ian e al., 2016). They a e de ined as in ec ions de eloped in a hospi al o o he heal h ca e acili y ha i s appea 48 h o mo e a e hospi al admission, o wi hin 30 days a e ha ing ecei ed heal h ca e (Haque e al., 2018). The e o e, he abili y o iden i y he sou ce o he in ec ion and moni o he sp ead o disease is essen ial. To each ha goal, a p ocess called bac e ial yping is used (Ruppi sch, 2016) as i can dis inguish indi idual bac e ial s ains om one species. Howe e , apid labo a o y and compu a ional echniques’ esolu ion o yping is limi ed and he abili y o dis inguish s ains among a low di e si y hospi al bac e ial popula ion is p ac ically impossible. Me hods such as pulsed- ield gel elec opho esis (PFGE) (Schwa z and Can o , 1984), epe i i e sequence-based PCR ( ep-PCR) (Sku ko a e al., 2019), mul ilocus sequence yping (MLST) (Maiden e al., 1998), o mini-MLST (Bezdicek e al., 2019) a e used o cha ac e ize and in es iga e ou b eaks. As housekeeping genes used in MLST o mini-MLST may no ha e su icien a iabili y in some cases, he e is s ill a need o de elop new app oaches o iden i y a iable sequences. One o hem is mul ispace yping (MST) which uses sequences occu ing in junk DNA (in e genic egions and pseudogenes) as hey show highe a iabili y han coding genome pa s (Foucaul e al., 2005). Ano he app oach is single-locus sequence yping (SLST), whe e one a iable sequence iden i ied based on a genome mining app oach can ha e simila disc imina o y powe o MLST (Scholz e al., 2014). Howe e , i p ecise Nyk yno a e al. Highly Disc imina i e Sequence Ma ke s De ec ion analyses o he low di e si ied bac e ial popula ion a e equi ed, he only me hod ha can be used is whole-genome sequencing (WGS). Fo WGS yping, wo main app oaches exis : single nucleo ide a ian analysis (SNV) and co e genome MLST (cgMLST) o whole genome MLST (wgMLST) (Hen i e al., 2017; Schü ch e al., 2018), whe e genes common o bac e ial isola es a e compa ed. Ne e heless, he yping me hods which equi e WGS a e no sui able o ou ine clinical p ac ice. Addi ionally, he compa ison o di e en s udies is di icul as di e en da a quali y assessmen , genome assembly, and esul s analysis a e done (Ruppi sch, 2016). In his a icle, we p esen a new app oach o localize highly a iable sequences in bac e ial genomes o Esche ichia coli and En e ococcus aecium. De ec ing new yping agmen s is based on a calcula ion o wo d en opy in genomes assembled om NGS da a. Po en ial a iable sequences a e e alua ed by phylogene ic analysis, and he ob ained esul s a e compa ed o cgMLST. The co e genome was chosen as he iden i ied ma ke s mus be p esen ed in all bac e ial s ains o gi en species. Fo ha eason, i is no app op ia e o s udy he pan-genome as i con ains he co e genome and he dispensable genome (Medini e al., 2005). The iden i ied ma ke s will be used o closely ela ed bac e ial yping ia s anda d labo a o y me hods wi hou he need o u he WGS. Ou app oach can be used o any bac e ia; he only equi emen is aligned isola e genomes. 2. MATERIALS AND METHODS 2.1. Da ase Fo his a icle, samples we e collec ed in he Uni e si y Hospi al o B no. In o al, 23 E. aecium isola es we e ob ained om se en hospi al depa men s be ween 6/2017 and 2/2018 and 21 E. coli isola es we e collec ed om a single hospi al depa men be ween 5/2019 and 7/2019. The e o e, he da ase s ha e di e en a iabili y a es as hey we e collec ed du ing di e en pe iods and om a di e en numbe o depa men s. Sequencing lib a ies we e p epa ed wi h KAPA Hype Plus Ki s (Roche, Swi ze land), and a quali y check was pe o med using a 2100 Bioanalyze (Agilen Technologies, USA). Lib a ies we e quan i ied wi h KAPA Lib a y Quan i ica ion ki (Roche, Swi ze land). Sequencing was pe o med on an Illumina MiSeq pla o m using MiSeq Reagen Ki 2 (500-cycles) (Illumina, USA) and pai ed-end eads (250 bp) we e acqui ed. 2.2. Re e ence-Based Genome Mapping A e he sequenced da a’s quali y con ol, he eads we e mapped ia BBMap ( 38.71, Bushnell, 2014) o he human genome (GRCh38.p13) o emo e con amina ion which could eme ge du ing sample p epa a ion. Reads which did no map o he human genome we e u he analyzed and quali y assessmen was done. Adap e s and low-quali y pa s o eads we e immed ia T immoma ic ( 0.36, Bolge e al., 2014). The eads leng h was abou 250 bp; hus he e e ence-based genome mapping app oach was chosen, and o his pu pose, he Bu ows-Wheele Alignmen MEM ( 0.7.17- 1188, Li, 2013) was used. The e e ence sequences we e ob ained om he GenBank da abase (Cla k e al., 2016). As e e ence sequence o assembly o E. aecium genomes was chosen CP003351.1 (Lam e al., 2012) and E. coli genomes we e assembled agains he sequence BA000007.3 (Hayashi, 2001). Then Sam ools ( 1.7, Li e al., 2009) was used o emo e low-quali y and duplica ed eads, and consensus sequences we e gene a ed. 2.3. The En opy-Based De ec ion o Va iable Sequences To loca e highly a iable egions in sequences, he en opy alue can be used. The DNA sequences en opy es ima ion (Schmi and He zel, 1997) is based on he Shannon-en opy (Shannon, 1948). In his pape , he en opy calcula ion modi ica ion as in e se alue o in o ma ion con en is u ilized. This modi ica ion commonly used o sequence logo de e mina ion (Schneide and S ephens, 1990) measu es he conse a ion o posi ion in bi s which be e espec s di e en cha se s wi h a iable con en o ambi alen bases. En opy Hi(Schneide and S ephens, 1990) can be calcula ed as Hi= − X a=A,C,G,T a,i·log2 a,i, (1) whe e iis he posi ion in he sequence, is he equency o one o ou nucleo ides’ occu ence (a= A, C, G, T) in posi ion iand can be calcula ed as a,i=na n, (2) whe e nais numbe o occu ences o one nucleo ide in one posi ion and nis numbe o all cha ac e s a he gi en posi ion. In he ideal case, we wan o ind such a egion capable o dis inguishing each isola e. Fo ha eason, he cu en me hod o en opy calcula ion in one posi ion (Nyk yno a e al., 2019) in he sequence’s alignmen canno be used, as i can only dis inguish ou a ian s because ou nucleo ides (A, C, G, T) a e p esen ed. In his a icle, a new unique app oach using wo d en opy is p esen ed, whe e he wo d is a nucleo ide sequence o de ined leng h. As e e ence-based genome assembly is no pe ec du ing consensus calling ambiguous nucleo ides can appea in he consensus sequences, and hey a i icially inc ease he en opy alue as shown in Figu e 1. To p e en de ec ing alsely a iable sequences, he whole columns o aligned sequences whe e ambiguous nucleo ides occu we e emo ed and o p ese e he o iginal leng h o he c ea ed consensus sequences eplaced by gaps. Also, he de ec ed ma ke s should occu in all analyzed isola es. Thus, when gaps appea in some sequences, he gaps ep esen ed by dashes a e added o he whole columns as is also shown in Figu e 1. In he window w, whose size co esponds o wo d leng h, he numbe o unique wo ds is de e mined, and hen he en opy o wo ds Hwis calcula ed as Hw= − X(kw n·(log2 kw n)), (3) F on ie s in Mic obiology | www. on ie sin.o g 2Feb ua y 2021 | Volume 12 | A icle 631605 Nyk yno a e al. Highly Disc imina i e Sequence Ma ke s De ec ion FIGURE 1 | Example o sequence p ep ocessing. (A) Example o aligned sequences be o e p ep ocessing and hei en opy o wo d alues o wo ds o leng h i e. (B) Aligned sequences a e p ep ocessing and co esponding en opy o wo d alues. FIGURE 2 | Scheme o en opy o wo ds’ calcula ion o a se o sequences a e p ep ocessing and o wo d leng h o 5 bp and one nucleo ide window shi . whe e kwis he numbe o occu ences o unique wo ds in each alignmen in he cu en window, and nis he numbe o isola es. The sum is applied o e he alignmen in he cu en window. A e es ima ing he en opy alue, he window mo es wi h a shi o one nucleo ide. The p inciple o en opy o wo ds’ calcula ion is shown in Figu e 2. Fo de ined sliding window sizes w(also called wo d leng h), he en opy signal wi h leng h N−w+1 is ob ained, whe e Nis he analyzed genome’s leng h. In he signal, he maximum alue is iden i ied. In heo y, he maximum alue can be de ined as a bina y loga i hm o he numbe o sequences. Based on he maximum alue F on ie s in Mic obiology | www. on ie sin.o g 3Feb ua y 2021 | Volume 12 | A icle 631605 Nyk yno a e al. Highly Disc imina i e Sequence Ma ke s De ec ion ha occu s in he signal, he h eshold is se as a maximum alue educed by 0.4 bi , and signal peaks abo e he h eshold a e labeled as po en ial gene ic ma ke s. The s a ed alue was chosen empi ically based on he numbe o po en ial a iable egions o ensu e ha some egions will ha e su icien disc imina o y powe ; howe e , i can be adjus ed o analyzed da a by he use . The aligned sequences co esponding o peaks a e ex ac ed and analyzed. I wo peaks a e dis an less han he window leng h, he sequences co esponding o he peaks a e me ged and analyzed as one. De eloped sou ce codes and unc ions in Ma lab R2020a can be ound a h ps://gi hub.com/ma ke anyk yno a/ma ke s_de ec ion. 2.4. Phylogene ic Analysis In he i s pa , each se o aligned sequences co esponding o peaks was analyzed. The e olu iona y dis ances we e calcula ed based on he Kimu a (Kimu a, 1980) model, which is used as s anda d. F om ob ained dis ances, he phylogene ic ees we e cons uc ed based on UPGMA. Clus e analysis was conduc ed, and he clus e s om he ees we e es ablished and compa ed o esul s ob ained om cgMLST. The h eshold o clus e s de e mina ion was se o in e sec he ee so ha he maximum numbe o clus e s wasob ained. The esul s showed ha no se had su icien disc imina o y powe o dis inguish be ween s ains. Tha is why he combina ion o wo aligned sequences se s was used. The iden i ied a iable se s o sequences we e me ged, which c ea ed one long sequence o each analyzed bac e ial isola e. Then o me ged sequences, again he e olu iona y dis ances we e calcula ed and based on hem, he phylogene ic ees we e cons uc ed and examined. The leng h o a a iable se o sequences had a p opo ional impac on he c ea ed phylogene ic ee. I he sequences in one a iable se had a longe leng h, hey had a highe impac on he ee and ice e sa, he se o sho e sequences did no a ec he inal ee so much. Ne e heless, sequences a iabili y has he main impac on he phylogene ic ee. 3. RESULTS AND DISCUSSION 3.1. Sequence Type De e mina ion and cgMLST Analysis The inco po a ed E. coli and E. aecium cgMLST, cgSNV, and wgSNV schemes in Ridom SeqSphe e+ so wa e (Ridom, DE) we e used o analyze he gene ic ela edness o sequenced bac e ial genomes. Fo E. coli, his included 3,152 a ge genes used o gene a e a cgMLST dend og am. Fo he cgSNV minimum spanning ee analysis, 152,212 aligned nucleo ide si es we e analyzed. In o al, 176,629 aligned nucleo ide si es we e used o wgSNV minimum spanning ee analysis. Fo E. aecium, he scheme included 1,423 a ge genes o cgMLST, 7,707 aligned nucleo ide si es o cgSNV and 10,858 aligned nucleo ide si es o wgSNV analysis. The WGS da a we e used o de e mine sequence ypes (STs) o bo h, E. coli and E. aecium s ains. Fo E. coli, se en housekeeping genes (adk, umC, gy B, lcd, mdh, pu A, ecA) and o E. aecium again se en housekeeping genes (a pA, ddl, gdh, pu K, gyd, ps S, adk) we e analyzed. Fo E. aecium h ee STs we e iden i ied (4 ×ST 117, 1 ×ST 18, 18 ×ST 80), and o E. coli ele en STs we e ecognized and one ST emains unknown (1 ×ST 69, 4 ×ST 131, 1 ×ST 95, 2 ×ST 404, 2 ×ST 38, 2 ×ST 1049, 4 ×ST 58, 1 ×ST 297, 1 ×ST 517, 2 ×ST 101, 1 ×ST UNW). The comple e esul s a e a ached as Supplemen a y File 1. 3.2. Po en ial Gene ic Ma ke s Analysis Sequences wi h a high alue o wo d en opy we e iden i ied o wo d leng hs om 50 o 400 bp wi h a s ep o 50 bp. T adi ional labo a o y me hods can use iden i ied ma ke s o leng h om his ange. In he i s pa , one a iable agmen wi h high wo d en opy alues was used o dis inguish he s ains o E. coli and E. aecium. Examining 1,358 agmen s wi h high a iabili y o all wo d sizes showed ha he maximum numbe o clus e s ha can be dis inguished o E. coli was 13. Fo E. aecium, 104 po en ial gene ic ma ke s o all wo d sizes we e analyzed, and he maximum numbe o dis inguishable clus e s was 5. The comple e esul s a e in Supplemen a y File 2. The numbe o dis inguishable clus e s using one a iable agmen is no high enough. Fo ha eason, wo a iable agmen s wi h wo d en opy abo e h eshold we e analyzed. Fo all combina ions o wo possible gene ic ma ke s, he phylogene ic ees we e cons uc ed, and he numbe o clus e s was calcula ed. In he nex s ep, he po en ial gene ic ma ke s wi h he highes numbe o clus e s we e analyzed. Fo all wo d sizes, 136,661 combina ions o possible gene ic ma ke s o E. coli and 685 o E. aecium we e examined in o al. In case o need, he combina ion o e en mo e ma ke s can be used; howe e , he ime o analysis will g ow as he ime complexi y is O(nm), whe e mis he numbe o a iable agmen s. Fo E. coli, 29 ees we e examined which di ided he isola es o 15 clus e s. Ten ees classi y he genomes acco ding o he 11 p ede ined clus e s om cgMLST. The esul s om cgMLST wi h 11 ma ked sequence ypes and 11 de ined clus e s and a cladog am c ea ed based on he combina ion o wo gene ic ma ke s wi h 15 labeled clus e s a e shown in Figu e 3. In each case o one gene ic ma ke , he egion which con ained he genes o hypo he ical p o ein (NP_308436.2) and S- o mylglu a hione hyd olase (NP_308437.1) was selec ed. In i e ou o en cases, he in e genic egion in on o hypo he ical p o ein was also inco po a ed. As he second gene ic ma ke , wo egions we e mos o en ( h ee imes) chosen. The i s one con ained ou e memb ane p o ein (NP_313225.1) and in one case he in e genic egion in on o he gene oo. The second one s a ed in imb ial-like adhesin p o ein (NP_310944.1) and ended in hypo he ical p o ein (NP_310945.1). Fou o he gene ic ma ke s we e selec ed only one ime. Two o hem consis ed o non-coding egion and gene [PhoH amily P-loop ATPase (NP_309293.1) in he i s case, excinuclease U ABC subuni U C (NP_310678.2) in he second case], o ano he ma ke pa o he gene alyl- RNA syn he ase (NP_313262.1) was picked, and in he las case, as a gene ic ma ke he non- coding egion was selec ed. De ailed in o ma ion is a ailable in F on ie s in Mic obiology | www. on ie sin.o g 4Feb ua y 2021 | Volume 12 | A icle 631605 Nyk yno a e al. Highly Disc imina i e Sequence Ma ke s De ec ion FIGURE 3 | Classi ica ion o 21 E. coli genomes. (A) Co e genome MLST analysis o 21 E. coli isola es wi h p ede ined ma ked clus e s (ou e ci cles) and sequence ypes (inne ci cles). (B) Cladog am o 21 E. coli isola es based on wo iden i ied gene ic ma ke s wi h highligh clus e s ob ained om co e genome MLST analysis (ou e ci cle) and subg oups ob ained based on analysis o iden i ied a iable sequences (inne ci cle) c ea ed by E ol iew (Sub amanian e al., 2019). FIGURE 4 | Classi ica ion o 23 E. aecium genomes. (A) Co e genome MLST analysis o 23 E. aecium isola es wi h p ede ined ma ked clus e s (ou e ci cles) and sequence ypes (inne ci cles). (B) Cladog am o 23 E. aecium isola es based on wo iden i ied gene ic ma ke s wi h highligh clus e s ob ained om co e genome MLST analysis (ou e ci cle) and subg oups ob ained based on analysis o iden i ied a iable sequences (inne ci cle) c ea ed by E ol iew (Sub amanian e al., 2019). Supplemen a y File 2 and all phylogene ic ees ha co ec ly classi ied genomes a e depic ed in Supplemen a y Figu e 1. Fo E. aecium, 22 ees which di ided he isola es in o nine clus e s we e examined. I was ound ha six ees classi y he genomes acco ding o i e p ede ined clus e s om cgMLST. In Figu e 4, he esul s om cgMLST analysis wi h h ee labeled sequence ypes, i e clus e s and a cladog am cons uc ed based on he iden i ied combina ion o wo gene ic ma ke s wi h nine labeled clus e s shown. As in he p e ious case, in all possible combina ions, one gene ic ma ke always emained he same, and i was RNA me hyl ans e ase (WP_002294889.1). In one o six cases, he in e genic egion a ound he gene was also inco po a ed in o he a iable gene ic pa . Mg- ansloca ing P- ype ATPase (WP_002304253.1) was in mos cases ( h ee imes) selec ed as he second ma ke . The egion ha con ained he gene o DUF1430 domain-con aining p o ein (WP_002288541.1) was chosen wice. Fo he las possible gene ic ma ke , he gene o nucleoside-diphospha e kinase (WP_002293327.1) was picked. The comple e esul s a e a ailable in Supplemen a y File 2 and all phylogene ic ees co ec ly classi ying genomes a e shown in Supplemen a y Figu e 2. 3.3. Wo d Sizes Fo de ined wo d sizes, a maximum numbe o co ec ly classi ied clus e s o phylogene ic ees based on wo ma ke combina ions F on ie s in Mic obiology | www. on ie sin.o g 5Feb ua y 2021 | Volume 12 | A icle 631605 Nyk yno a e al. Highly Disc imina i e Sequence Ma ke s De ec ion FIGURE 5 | Analysis o wo d size impac on numbe o co ec ly classi ied clus e s. (A) Maximum numbe o co ec ly classi ied clus e s o combina ions o wo a iable agmen s o di e en wo d sizes and numbe o wo a iable agmen combina ions ha co ec ed classi ied analyzed E. aecium isola es. (B) Maximum numbe o co ec ly classi ied clus e s o combina ion o wo a iable agmen s o di e en wo d sizes and numbe o wo a iable agmen combina ions ha co ec ed classi ied analyzed E. coli isola es. was es ablished, as is shown in Figu e 5. Fo analyzed ees he numbe o clus e s was de e mined and he co ec genomes classi ica ion was con olled. The esul s showed ha o E. aecium he wo d leng hs do no signi ican ly a ec he maximum numbe o co ec ly es ablished clus e s, which is nine o all wo d sizes excep he leng h 50 and 300 bp. On he con a y, he esolu ion powe o E. coli g ows wi h inc easing wo d leng hs and he bes esul s (dis inguish 15 clus e s) can be ob ained o a window o size 300 bp and mo e. 3.4. Resul s Ve i ica ion The analyzed da ase s we e ex ended by comple e genomes sequences o E. coli and E. aecium ob ained om he GenBank da abase. Fo each bac e ium, six sequences we e downloaded, and he STs we e de e mined. Fo E. coli isola es (CP050212.1, CP050219.1, CP050211.1, CP050201.1, CP050214.1, CP050205.1) i e STs we e iden i ied (1 ×ST 2705, 1×ST 2280, 1 ×ST 70, 1 ×ST 602, 2 ×ST 131), and o E. aecium isola es (CP040236.1, CP036151.1, CP027501.1, CP035666.1, CP035660.1, CP035220.1) one ST was ecognized, and one ST emains unknown (5 ×ST 80, 1 ×ST UNW). The esul s a e in Supplemen a y File 1. Using he BLAST+ ( 2.6.0+, Camacho e al., 2009) he iden i ied a iable ma ke s we e loca ed in hese genomes. The ma ke s sequences o bo h se s (o iginal and da abase) we e analyzed oge he . The sequences we e aligned, e olu iona y dis ances we e calcula ed, and phylogene ic ees we e cons uc ed. To ally, 27 isola es o E. coli belonging o nine STs we e classi ied o 17 clus e s and 29 isola es o E. aecium belonging o ou STs we e i e imes classi ied o 11 clus e s and once o 10 clus e s. All c ea ed phylogene ic ees a e depic ed in Supplemen a y Figu e 3. The iden i ied gene ic ma ke s can be used o yping no el isola es. Howe e , i a la ge numbe o new isola es should be analyzed, he algo i hm should be un again o ind ma ke s wi h he highes disc imina o y powe o he analyzed bac e ial popula ion. 3.5. Analyzed Bac e ia The p oposed app oach was es ed on wo bac e ia—E. aecium and E. coli which belong among signi ican HAIs pa hogens. E. aecium is a G am-posi i e bac e ium and can be a sou ce o nosocomial in ec ions, especially in immunocomp omised pa ien s, and causes endoca di is, bac e emia o u ina y ac in ec ion. In addi ion, some E. aecium isola es can de elop d ug esis ance o se e al an ibio ics g oups such as glycopep ides, be a-lac ams, luo oquinolones, o aminoglycosides (Yoong e al., 2004; Cas illo-Rojas e al., 2013). Ano he pa hogen om G am-nega i e bac e ia is E. coli, which can be ound in human in es inal lo a as a commensal. Howe e , as an oppo unis ic pa hogen E. coli ep esen s a huge public heal h p oblem. I is one o he mos common causes o HAIs ypically connec ed o meningi is, pneumonia, u ina y ac in ec ions, and so issue in ec ions (Jau eguy e al., 2008). E. coli s ains ha e he abili y o accumula e esis ance genes; hus, he esis ance o b oad-spec um cephalospo ins, ca bapenems, aminoglycoside, luo oquinolones, and polymyxins is o en obse ed (Poi el e al., 2018). F on ie s in Mic obiology | www. on ie sin.o g 6Feb ua y 2021 | Volume 12 | A icle 631605 Nyk yno a e al. Highly Disc imina i e Sequence Ma ke s De ec ion 4. CONCLUSION As yping is c ucial o iden i ying he sou ce o in ec ion and ou b eak moni o ing, i is necessa y o ha e a highly sensi i e yping me hod o dis inguish closely ela ed bac e ial s ains. Howe e , cu en ly used me hods ha e insu icien disc imina o y powe excep whole genome sequencing. Ne e heless, WGS is ime-consuming, and da a p ocessing is no possible in ou ine clinical p ac ice. Fo his eason, a new app oach o de ec a iable gene ic ma ke s om WGS da a is p esen ed. The WGS is only used once o ob ain he i s se o da a o gene ic ma ke s de ec ion. To loca e highly a iable sequences, wo d en opy is employed. The loca ed ma ke s can be used o yping a closely ela ed bac e ial popula ion ia s anda d labo a o y me hods. Thus, ano he genome sequencing will be necessa y only i a la ge numbe o new isola es should be analyzed. Using he en opy-based app oach, new gene ic ma ke s wi h he same o e en highe disc imina o y powe han cgMLST can be iden i ied. As he leng h o wo ds om 50 o 400 bp wi h a s ep o 50 bp is applied, po en ial ma ke s can be used in s anda d labo a o y echniques because hey ha e sui able leng h. The only equi emen o he p oposed app oach is ha genomes mus be aligned, which can cause a p oblem mainly when la ge genomes should be examined. The p esen ed me hod was es ed on bo h G am-posi i e and G am-nega i e bac e ia. As in bo h es ed da ase s, one ma ke wi h su icien disc imina o y powe o yping o low di e si y bac e ial popula ion was no ound, he combina ion o wo ma ke s was employed. Based on wo iden i ied ma ke s, we we e able o dis inguish he G am-nega i e and G am-posi i e bac e ial popula ion a he same and e en a highe disc imina ion le el han ia classic labo a o y me hods using only wo a iable sequences ins ead o se en housekeeping genes. The iden i ied ma ke s will be used in ou ine labo a o y me hods o bac e ial yping. The low numbe o a iable agmen s ha should be examined will sho en he ime o analysis. Commonly used yping me hods mainly use coding egions and analyze genes conse ed and p esen ed in all isola es o he gi en species. Hence he a iabili y a e is lowe and yping he closely ela ed popula ion is no possible. The p oposed me hod uses he whole genome ange; hus, also in e genic genome egions a e analyzed as hey ha e highe a iabili y han coding a eas and can con ain gene ic ma ke s o yping ela ed isola es. Based on iden i ied ma ke s, he closely ela ed bac e ial popula ion can be di e si ied and o ind ou he ansmission ou es and sou ce o ou b eaks will be possible. DATA AVAILABILITY STATEMENT The da ase s p esen ed in his s udy can be ound in online eposi o ies. The names o he eposi o y/ eposi o ies and accession numbe (s) can be ound a : h ps://www.ncbi.nlm.nih. go /, PRJNA675431. AUTHOR CONTRIBUTIONS MN, VB, KS, and HS con ibu ed o he concep ion and design o he s udy. MN implemen ed he algo i hm and e alua ed he esul s. ML and MB ensu ed he biological aspec s o he p ojec . All au ho s ead and app o ed he inal manusc ip . FUNDING This wo k was suppo ed by a g an p ojec om he Czech Science Founda ion (GA17-01821S). SUPPLEMENTARY MATERIAL The Supplemen a y Ma e ial o his a icle can be ound online a : h ps://www. on ie sin.o g/a icles/10.3389/ micb. 2021.631605/ ull#supplemen a y-ma e ial Supplemen a y File 1 | MLST analysis esul s o E.coli and E. aecium. Supplemen a y File 2 | Po en ial and inal gene ic ma ke s o E. coli and E. aecium. Supplemen a y Figu e 1 | Phylogene ic ees ha co ec ly classi ied E. coli genomes wi h ma ked clus e s. Supplemen a y Figu e 2 | Phylogene ic ees ha co ec ly classi ied E. aecium genomes wi h ma ked clus e s. Supplemen a y Figu e 3 | Phylogene ic ees o o iginal and da abase genomes o E. aecium and E. coli genomes wi h ma ked clus e s. REFERENCES A e ian, H., Vogel, M., Kwe ka , A., and Ha mann, M. (2016). Economic e alua ion o in e en ions o p e en ion o hospi al acqui ed in ec ions: a sys ema ic e iew. PLoS ONE 11:e0146381. doi: 10.1371/jou nal.pone.0146381 Bezdicek, M., Nyk yno a, M., Ple o a, K., B helo a, E., Kocmano a, I., Sedla , K., e al. (2019). Applica ion o mini-MLST and whole genome sequencing in low di e si y hospi al ex ended-spec um be a-lac amase p oducing Klebsiella pneumoniae popula ion. PLoS ONE 14:e0221187. doi: 10.1371/jou nal.pone.0221187 Bolge , A. M., Lohse, M., and Usadel, B. (2014). T immoma ic: a lexible imme o Illumina sequence da a. Bioin o ma ics 30, 2114–2120. doi: 10.1093/bioin o ma ics/b u170 Bushnell, B. (2014). BBMap: A Fas , Accu a e, Splice-Awa e Aligne . No. LBNL- 7065E. Be keley, CA: E nes O lando Law ence Be keley Na ional Labo a o y. Camacho, C., Coulou is, G., A agyan, V., Ma, N., Papadopoulos, J., Beale , K., e al. (2009). BLAST+: a chi ec u e and applica ions. BMC Bioin o ma ics 10:421. doi: 10.1186/1471-2105-10-421 Cas illo-Rojas, G., Maza i-Hi ía , M., Ponce de León, S., Amie a-Fe nández, R. I., Agis-Juá ez, R. A., Huebne , J., e al. (2013). Compa ison o En e ococcus aecium and En e ococcus aecalis s ains isola ed om wa e and clinical samples: an imic obial suscep ibili y and gene ic ela ionships. PLoS ONE 8:e59491. doi: 10.1371/jou nal.pone. 0059491 Cla k, K., Ka sch-Miz achi, I., Lipman, D. J., Os ell, J., and Saye s, E. W. (2016). GenBank. Nucleic Acids Res. 44, D67–D72. doi: 10.1093/na /gk 1276 Foucaul , C., La Scola, B., Lind oos, H., Ande sson, S. G. E., and Raoul , D. (2005). Mul ispace yping echnique o sequence-based yping o Ba onella quin ana. J. Clin. Mic obiol. 43, 41–48. doi: 10.1128/JCM.43.1.41-4 8.2005 F on ie s in Mic obiology | www. on ie sin.o g 7Feb ua y 2021 | Volume 12 | A icle 631605 Nyk yno a e al. Highly Disc imina i e Sequence Ma ke s De ec ion Haque, M., Sa elli, M., McKimm, J., and Abu Baka , M. B. (2018). Heal h ca e-associa ed in ec ions-an o e iew. In ec . D ug Resis . 11, 2321–2333. doi: 10.2147/IDR.S177247 Hayashi, T. (2001). Comple e genome sequence o en e ohemo hagic Eschelichia coli O157:H7 and genomic compa ison wi h a labo a o y s ain K-12. DNA Res. 8, 11–22. doi: 10.1093/dna es/8.1.11 Hen i, C., Leeki cha oenphon, P., Ca le on, H. A., Radomski, N., Kaas, R. S., Ma ie , J.-F., e al. (2017). An assessmen o di e en genomic app oaches o in e ing phylogeny o Lis e ia monocy ogenes. F on . Mic obiol. 8:2351. doi: 10.3389/ micb.2017.02351 Jau eguy, F., Land aud, L., Passe , V., Diancou , L., F apy, E., Guigon, G., e al. (2008). Phylogene ic and genomic di e si y o human bac e emic Esche ichia coli s ains. BMC Genomics 9:560. doi: 10.1186/1471-2164-9-560 Kimu a, M. (1980). A simple me hod o es ima ing e olu iona y a es o base subs i u ions h ough compa a i e s udies o nucleo ide sequences. J. Mol. E ol. 16, 111–120. doi: 10.1007/BF01731581 Lam, M. M. C., Seemann, T., Bulach, D. M., Gladman, S. L., Chen, H., Ha ing, V., e al. (2012). Compa a i e analysis o he i s comple e En e ococcus aecium genome. J. Bac e iol. 194, 2334–2341. doi: 10.1128/JB.00259-12 Li, H. (2013). Aligning sequence eads, clone sequences and assembly con igs wi h BWA-MEM. a Xi [P ep in ]. a Xi :1303.3997 Li, H., Handsake , B., Wysoke , A., Fennell, T., Ruan, J., Home , N., e al. (2009). The Sequence Alignmen /Map o ma and SAM ools. Bioin o ma ics 25, 2078–2079. doi: 10.1093/bioin o ma ics/b p352 Maiden, M. C. J., Byg a es, J. A., Feil, E., Mo elli, G., Russell, J. E., U win, R., e al. (1998). Mul ilocus sequence yping: a po able app oach o he iden i ica ion o clones wi hin popula ions o pa hogenic mic oo ganisms. P oc. Na l. Acad. Sci. U.S.A. 95, 3140–3145. doi: 10.1073/pnas.95.6.3140 Medini, D., Dona i, C., Te elin, H., Masignani, V., and Rappuoli, R. (2005). The mic obial pan-genome. Cu . Opin. Gene . De . 15, 589–594. doi: 10.1016/j.gde.2005.09.006 Nyk yno a, M., Made anko a, D., Ba on, V., Bezdicek, M., Lenge o a, M., and Sku ko a, H. (2019). “En opy-based de ec ion o gene ic ma ke s o bac e ia geno yping,” in Lec u e No es in Compu e Science (including subse ies Lec u e No es in A i icial In elligence and Lec u e No es in Bioin o ma ics) (Cham: Sp inge ), 11466 LNBI, 177–188. doi: 10.1007/978-3-030-17935-9_17 Poi el, L., Madec, J.-Y., Lupo, A., Schink, A.-K., Kie e , N., No dmann, P., e al. (2018). “An imic obial esis ance in Esche ichia coli,” in An imic obial Resis ance in Bac e ia om Li es ock and Companion Animals, eds S. Schwa z, L. M. Ca aco, and J. Shen (Washing on, DC: Ame ican Socie y o Mic obiology), 289–316. doi: 10.1128/mic obiolspec.ARBA-0026-2017 Ruppi sch, W. (2016). Molecula yping o bac e ia o epidemiological su eillance and ou b eak in es iga ion/Molekula e Typisie ung on Bak e ien ü die epidemiologische Übe wachung und Ausb uchsabklä ung. J. Land Manage. Food En i on. 67, 199–224. doi: 10.1515/boku-20 16-0017 Schmi , A. O., and He zel, H. (1997). Es ima ing he en opy o DNA sequences. J. Theo . Biol. 188, 369–377. doi: 10.1006/j bi.1997.0493 Schneide , T. D., and S ephens, R. M. (1990). Sequence logos: a new way o display consensus sequences. Nucl. Acids Res. 18, 6097–6100. doi: 10.1093/na /18.20.6097 Scholz, C. F. P., Jensen, A., Lomhol , H. B., B üggemann, H., and Kilian, M. (2014). A no el high- esolu ion single locus sequence yping scheme o mixed popula ions o P opionibac e ium acnes in i o. PLoS ONE 9:e104199. doi: 10.1371/jou nal.pone.0104199 Schü ch, A., A edondo-Alonso, S., Willems, R., and Goe ing, R. (2018). Whole genome sequencing op ions o bac e ial s ain yping and epidemiologic analysis based on single nucleo ide polymo phism e sus gene-by-gene-based app oaches. Clin. Mic obiol. In ec . 24, 350–354. doi: 10.1016/j.cmi.2017.12.016 Schwa z, D. C., and Can o , C. R. (1984). Sepa a ion o yeas ch omosome- sized DNAs by pulsed ield g adien gel elec opho esis. Cell 37, 67–75. doi: 10.1016/0092-8674(84)90301-5 Shannon, C. E. (1948). A ma hema ical heo y o communica ion. Bell Sys . Techn. J. 27, 379–423. doi: 10.1002/j.1538-7305.1948. b01338.x Sku ko a, H., Vi ek, M., Bezdicek, M., B helo a, E., and Lenge o a, M. (2019). Ad anced DNA inge p in geno yping based on a model de eloped om eal chip elec opho esis da a. J. Ad . Res. 18, 9–18. doi: 10.1016/j.ja e.2019.01.005 Sub amanian, B., Gao, S., Le che , M. J., Hu, S., and Chen, W.-H. (2019). E ol iew 3: a webse e o isualiza ion, anno a ion, and managemen o phylogene ic ees. Nucl. Acids Res. 47, W270-W275. doi: 10.1093/na /gkz357 Yoong, P., Schuch, R., Nelson, D., and Fische i, V. A. (2004). Iden i ica ion o a b oadly ac i e phage ly ic enzyme wi h le hal ac i i y agains an ibio ic- esis an En e ococcus aecalis and En e ococcus aecium. J. Bac e iol. 186, 4808–4812. doi: 10.1128/JB.186.14.4808-4812.2004 Con lic o In e es : The au ho s decla e ha he esea ch was conduc ed in he absence o any comme cial o inancial ela ionships ha could be cons ued as a po en ial con lic o in e es . Copy igh © 2021 Nyk yno a, Ba on, Sedla , Bezdicek, Lenge o a and Sku ko a. This is an open-access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion License (CC BY). The use, dis ibu ion o ep oduc ion in o he o ums is pe mi ed, p o ided he o iginal au ho (s) and he copy igh owne (s) a e c edi ed and ha he o iginal publica ion in his jou nal is ci ed, in acco dance wi h accep ed academic p ac ice. No use, dis ibu ion o ep oduc ion is pe mi ed which does no comply wi h hese e ms. F on ie s in Mic obiology | www. on ie sin.o g 8Feb ua y 2021 | Volume 12 | A icle 631605