FireProt: web server for automated design of thermostable proteins

Published online 26 Ap il 2017 Nucleic Acids Resea ch, 2017, Vol. 45, Web Se e issue W393–W399 doi: 10.1093/na /gkx285 Fi eP o : web se e o au oma ed design o he mos able p o eins Milos Musil1,2,3,†, Jan S ou ac1,3,†, Ja osla Bendl1,2,3, Jan B ezo sky1,3, Zbynek P okop1,3, Ja osla Zendulka2,4, Tomas Ma inek1,2,4, Da id Bedna 1,3,* and Ji i Dambo sky1,3,* 1Loschmid Labo a o ies, Depa men o Expe imen al Biology, Masa yk Uni e si y, B no, Czech Republic, 2Depa men o In o ma ion Sys ems, Facul y o In o ma ion Technology, B no Uni e si y o Technology, B no, Czech Republic, 3In e na ional Cen e o Clinical Resea ch, S . Anne’s Uni e si y Hospi al B no, B no, Czech Republic and 4Cen e o Excellence IT4Inno a ions, Technical Uni e si y Os a a, Os a a Recei ed Feb ua y 11, 2017; Re ised Ap il 02, 2017; Edi o ial Decision Ap il 10, 2017; Accep ed Ap il 11, 2017 ABSTRACT The e is a con inuous in e es in inc easing p o- eins s abili y o enhance hei usabili y in nume - ous biomedical and bio echnological applica ions. A numbe o in silico ools o he p edic ion o he e ec o mu a ions on p o ein s abili y ha e been de- eloped ecen ly. Howe e , only single-poin mu a- ions wi h a small e ec on p o ein s abili y a e ypi- cally p edic ed wi h he exis ing ools and ha e o be ollowed by labo ious p o ein exp ession, pu i ica- ion, and cha ac e iza ion. He e, we p esen Fi eP o , a web se e o he au oma ed design o mul iple- poin he mos able mu an p o eins ha combines s uc u al and e olu iona y in o ma ion in i s calcu- la ion co e. Fi eP o u ilizes six een ools and h ee p o ein enginee ing s a egies o making eliable p o ein designs. The se e is complemen ed wi h in e ac i e, easy- o-use in e ace ha allows use s o di ec ly analyze and op ionally modi y designed he mos able mu an s. Fi eP o is eely a ailable a h p://loschmid .chemi.muni.cz/ i ep o . INTRODUCTION P o eins a e widely used in nume ous biomedical and bio echnological applica ions. Howe e , na u ally occu - ing p o eins canno usually wi hs and he ha sh indus ial en i onmen , since hey a e mos ly e ol ed o unc ion a mild condi ions (1). P o ein enginee ing has e olu ionized he u iliza ion o na u ally a ailable p o eins o di e en indus ial applica ions by imp o ing a ious p o ein ea- u es such as s abili y, ac i i y o enan ioselec i i y o su - pass hei na u al limi a ions. P o ein s abili y is gene ally s ongly co ela ed wi h i s exp ession yield (2), hal -li e (3), se um su i al ime (4) and pe o mance in he p esence o dena u ing agen s (5). Thus, s abili y is one o he key de- e minan s o p o eins applicabili y in bio echnological p o- cesses. In he ideal case, he sa u a ion mu agenesis would be applied o e alua e e e y possible mu a ion on e e y posi- ion o he enginee ed p o ein (6). Howe e , such a sea ch space would be eno mous and he expe imen al e alua- ion can delay he design o uly he mos able p o ein o mon hs o e en yea s. The e o e, he e a e demands o e - ec i e and p ecise p edic i e compu a ion o p o ein s a- bili y. To sa is y his goal a numbe o in silico ools ha e been de eloped ecen ly. Some o hese ools such as EASE- MM (7), I-Mu an (8)o mCSM(9)a ebasedonma- chine lea ning echniques. O he s a e using so-called ene - ge ic unc ions. These p og ams can be u he ca ego ized in o wo g oups. The i s g oup u ilizes a physical e ec i e ene gy unc ion o simula ing he undamen al o ces be- ween a oms and is ep esen ed by he p og ams like Rose a (10) and E is (11). The second g oup is based on s a is ical po en ials o which he ene gies a e de i ed om equen- cies o esidues o a om con ac s epo ed in he da ase s o expe imen ally cha ac e ized p o ein mu an s, e.g. Pop- MuSiC (12) and FoldX (13). Howe e , due o he po en- ially an agonis ic e ec o mu a ions, only single-poin mu- a ions a e usually p edic ed in silico andha e obe ol- lowed by labo ious and cos ly p o ein exp ession, pu i ica- ion and cha ac e iza ion. Single-poin mu a ions ypically enhance he mel ing empe a u e o a ge p o eins by uni s o deg ee (3,14). A much highe deg ee o s abiliza ion can be achie ed by cons uc ing mul iple-poin mu an s (15). We ha e ecen ly de eloped he Fi eP o (16), combining ene gy- and e olu ion-based app oaches o eliable design o s able mul iple-poin mu an s. The p o ocol includes se - e al p eceding il e s ha accele a e he calcula ion by omi - ing po en ially dele e ious mu a ions. Fi eP o is cu en ly *To whom co espondence should be add essed. Tel: +420 5 4949 3467; Fax: +420 5 4949 6302; Email: [email p o ec ed]uni.cz; Websi e add ess: h p:// loschmid .chemi.muni.cz/ i ep o ( ype o web se e : compu a ional wo k low) Co espondence may also be add essed o Da id Bedna . Email: [email p o ec ed]uni.cz †These au ho s con ibu ed equally o he wo k as i s au ho s. C The Au ho (s) 2017. Published by Ox o d Uni e si y P ess on behal o Nucleic Acids Resea ch. 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 (h p://c ea i ecommons.o g/licenses/by-nc/4.0/), which pe mi s non-comme cial e-use, dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal wo k is p ope ly ci ed. Fo comme cial e-use, please con ac jou nals.pe [email protected] W394 Nucleic Acids Resea ch, 2017, Vol. 45, Web Se e issue a ailable only in a s and-alone o ma and equi es ex en- si e expe ience in bioin o ma ics o ca y ou all necessa y s eps o he wo k low. Cu en ly, we a e awa e o only one se e o design o s able mul iple-poin mu an s - PROSS (17), u ilizing Rose a modeling and phylogene ic sequence in o ma ion in i s compu a ion co e. He e, we p esen a web e sion o Fi eP o o he au o- ma ed design o he mos able p o eins. Fi eP o in eg a es six een compu a ional ools and u ilizes bo h sequence and s uc u al in o ma ion. Fi eP o web se e p o ides use s wi h he mos able p o eins, cons uc ed by h ee dis inc s a egies: (i) e olu ion-based app oach, u ilizing back- o- consensus analysis; (ii) ene gy-based app oach, e alua ing change in ee ene gy upon mu a ion and (iii) combina ion o bo h e olu ion-based and ene gy-based app oaches. In ou iew, i is e y impo an o ha e his in eg a ed ap- p oach, since phylogene ic analysis enables iden i ica ion o he mu a ions s abilized by en opy, which canno be p e- dic ed by o ce ield calcula ions (Bee ens e al., unde e- iew). The se e allows use s o include p e e ed mu a- ions in o he he mos able p o ein, o gene a e co espond- ing s uc u es and sequences o gene syn heses. Compa ed o he p e iously published Fi eP o p o ocol (16), mini- mum e o and no bioin o ma ics knowledge is equi ed om use s o calcula e and analyze he esul s. Fu he - mo e, all inpu pa ame e s and compu a ional p o ocols we e op imized o minimize o he wise highly ime demand- ing p ocedu e. The se e was complemen ed wi h a g aph- ical in e ace allowing use s o di ec ly analyze he p o ein o in e es and design mul iple-poin mu an s. MATERIALS AND METHODS The basic wo k low o Fi eP o s a egy is ou lined in Fig- u e 1. In o de o design a highly eliable he mos able mul iple-poin mu an , a p o ein de ined by he use is an- no a ed using se e al p edic ion ools and da abases (Phase 1). Wi h his knowledge in hand, ene gy- and e olu ion- based app oach is applied o assemble a lis o po en ially s abilizing single-poin mu a ions (Phase 2). Finally, h ee mul iple-poin mu an s a e gene a ed in an addi i e man- ne , while emo ing po en ially an agonis ic e ec s o mu- a ions (Phase 3). Phase 1: Anno a ion o he p o ein Ini ially, he use is eques ed o speci y he p o ein s uc- u e, ei he by p o iding i s PDB ID o by uploading a use - de ined PDB ile. The biological assembly o he a ge p o- ein is hen au oma ically gene a ed by he MakeMul ime ool (h p://wa cu .uwa e loo.ca/ ools/makemul ime /). Se- quence homologs a e ob ained by pe o ming a BLAST sea ch (18) agains he UniRe 90 da abase (19), using he a ge p o ein sequence as an inpu que y. Iden i ied ho- mologs a e hen aligned wi h he que y p o ein using USE- ARCH (20), while sequences whose iden i y wi h he que y is below o abo e he use de ined h esholds (de aul : 30 and 90%) a e excluded om he lis o homologs. The emaining sequences a e clus e ed using UCLUST (20), wi h a 90% iden i y h eshold o emo e close homologs. The clus e ep esen a i es a e so ed based on he BLAST que y co e age and by de aul , he i s 200 o hem a e used o c ea e a mul iple sequence alignmen wi h Clus al Omega ool (21). The mul iple sequence alignmen is used o: (i) es ima e he conse a ion coe icien o each esidue posi- ion in he p o ein based on he Jensen–Shannon en opy (22); (ii) iden i y co ela ed posi ions employing a consen- sual decision o he OMES (23), MI (24), aMIc (25), DCA (26), SCA (27), ELSC (28), McBASC (29) and (iii) analyze amino acid equencies a indi idual posi ions wi hin he p o ein. Phase 2: P edic ion o single-poin mu a ions In acco dance wi h he o iginal Fi eP o p o ocol, po en- ially s abilizing single-poin mu a ions a e iden i ied ia wo sepa a e b anches: one elying on he es ima ion o he change o ee ene gy upon mu a ion and second u ilizing back- o-consensus app oach. The i s , ene gy-based app oach is employing FoldX and Rose a ools ha pe o med bes on ou es ing da ase . P eceding il e s accele a e he calcula ion by omi ing po- en ially dele e ious mu a ions. P io o he iden i ica ion o he single-poin mu a ions i sel , he a ge p o ein s uc u e is amended and minimized. FoldX p o ocol is u ilized o ill in he missing a oms in he esidues and pa ched s uc u e is consequen ly minimized wi h Rose a minimiza ion mod- ule. Conse ed and co ela ed posi ions a e immedia ely ex- cluded om u he analysis. I was obse ed ha unc- ional and s uc u al cons ain s in p o eins gene ally lead o he conse a ion o amino acid esidues (30–33). Simi- la ly, co ela ed esidues o dina ily help o main ain p o- ein unc ion, olding o s abili y (34–36). Mu a ions con- duc ed on hese posi ions a e he e o e conside ed unsa e by cu en Fi eP o s a egy, e en hough he e is ce ainly a space o mo e sophis ica ed ea men o co ela ed posi- ions, which will be u he de eloped in u u e e sions o Fi eP o se e . The emaining posi ions a e subjec ed o sa u a ion mu- agenesis by using FoldX ool. Mu a ions wi h p edic ed ddG o e gi en h eshold (de aul : –1 kcal/mol) a e s ee ed away and es is o wa ded o Rose a calcula ions. Finally, he mu a ions p edic ed by Rose a as s ongly s abilizing (de aul cu -o : –1 kcal/mol) a e agged as po en ial candi- da es o he design o he mul iple-poin mu an s. A high ime demands o Rose a analysis we e one o he mos exc ucia ing issues wi h he o iginal Fi eP o p o- ocol. E en wi h he applica ion o il e s o e 100 mu a- ions was usually le o p ecise, bu slow, Rose a calcula- ions. Fo his eason, we ha e e alua ed se e al o ce ields and Rose a p o ocols wi h he newly assembled da ase con aining 1573 mu a ions om P oThe m da abase (37) and Ho MuSiC da ase (38). Based on he esul s o he e alua ions, he bes ade-o be ween he ime equi e- men s and p ecision was selec ed. Wi h Rose a p o ocol 3, we ha e achie ed mo e han en old inc ease in calcula ion speed while p ese ing high p edic ion accu acy. De ails on da ase cons uc ion and p o ocols e alua ion can be ound in he Supplemen 1 (Supplemen a y Tables S1–S5). The second app oach is based on he in o ma ion ob- ained om mul iple sequence alignmen . The mos com- mon amino acid in each posi ion o p o ein sequence o en Nucleic Acids Resea ch, 2017, Vol. 45, Web Se e issue W395 Figu e 1. Wo k low o Fi eP o s a egy. p o ides a non-negligible e ec on p o ein s abili y (39–42). The e o e, Fi eP o implemen s majo i y and equency a- io app oach o iden i y mu a ions a posi ions whe e he wild- ype amino acid di e s om he mos p e alen one. By de aul , he single ou mu a ions a e loca ed in he posi- ions whe e he consensus esidue is p esen in a leas 50% o all analyzed sequences (majo i y me hod) o whe e con- sensus esidue equency is 40% and is a leas i e imes mo e equen han he wild- ype amino acid ( equency a- io me hod). These h esholds we e chosen in acco dance o he p e iously published Ho Spo Wiza d me hod (43). Se- lec ed mu a ions a e e alua ed by FoldX and he s abilizing ones a e lis ed as candida e mu a ions o he enginee ing o mul iple-poin mu an . Phase 3: Design o he mos able p o ein In o al, h ee p o ein designs a e p o ided by Fi eP o s a egy. The i s design includes only he mu a ions om ene gy-based app oach, he second con ains he mu a ions sugges ed by he e olu ion-based app oach and he hi d is he combina ion o bo h. Na u ally, because o po en ially an agonis ic e ec s be ween indi idual mu a ions, we can- no combine indi idual mu a ions blindly. To a oid possible clashes, Fi eP o s a egy is ying o minimize an agonis ic e ec s by u ilizing Rose a. In he i s s ep, all pai s o single-poin mu a ions wi hin he ange o 10 ˚ A a e e alua ed sepa a ely o ene gy- and e olu ion- based app oach. Once change in ee ene gy is ob ained o all esidue pai s, Fi eP o s a s o in oduce hem in o he mul iple-poin mu an in he o de based on hei p edic ed W396 Nucleic Acids Resea ch, 2017, Vol. 45, Web Se e issue s abili y, excluding he mu a ions ha a e colliding wi h al- eady included mu a ions. Algo i hm s ops once he e a e no mu a ions le o he s abilizing e ec o analyzed pai d ops below de ined h eshold. Upon he comple ion o p e ious s ep, p ocedu e is e- pea ed his ime conside ing only he pai s be ween he mu a ions chosen o he cons uc ion o ene gy- and e olu ion-based mu an s. Finally, s uc u es o all h ee mu- an s a e modeled using he Rose a p o ocol 16. DESCRIPTION OF THE WEB SERVER Inpu The only equi ed inpu o he web se e is a e ia y s uc- u e o he p o ein o in e es , p o ided ei he as a PDB ID o a use -de ined PDB ile. The use can hen choose a p e- de ined biological uni gene a ed by he MakeMul ime ool o manually selec chains o which he calcula ion should be pe o med. The calcula ions can be con igu ed in ei he basic o ad anced mode. In he basic mode, use is allowed o change he se ing o BLAST sea ch and alignmen cons uc ion. The ad anced mode expands he lis o modi iable pa ame e s by he ones connec ed wi h: (i) he iden i ica ion o consensus esidues by majo i y and equency a io app oach, (ii) he h esh- olds used by FoldX and Rose a p edic ion ools and (iii) he decision h eshold employed in he consensual analysis o co ela ed posi ions. Ad anced mode allows expe use s o ine- une he pa ame e s o calcula ion acco ding o s ud- ied sys ems. Howe e , he p esen ed de aul alues a e op- imized o p o ide eliable esul s o mos o he sys ems and we he e o e do no ad ice hei change in he gene al scena ios. Ou pu Upon submission, a unique iden i ie is assigned o each job o ack he calcula ion and he ‘Resul s b owse ’ in o ms he use abou he s a us o he indi idual s eps in he Fi e- P o wo k low (Figu e 2B). Once he job is inished, use s can ei he di ec ly download he esul s in he .zip a chi e o na iga e hemsel es in o he ‘Resul s page’ o u he anal- ysis. The ‘Resul s page’ is in ui i ely o ganized in o se e al panels as desc ibed below. P o ein isualiza ion. The wild- ype and he mu an s uc- u e is in e ac i ely isualized in he web b owse (Fig- u e 2D) u ilizing he Jsmol apple (h p://wiki.jmol.o g/ index.php/JSmol). Use s can swi ch be ween di e en p o- ein isualiza ion s yles and also highligh selec ed amino acids in he p o ein s uc u e. Residues ha we e included in o ene gy-based mu an a e colo ed in o ange, e olu ion- based mu a ions a e in blue and all o he esidues a e in g ay. Use selec ed esidues ha we e no pa o any mu- an a e unde lined in ed. Mu an o e iew. The ‘Mu an o e iew’ panel is o ga- nized in o ou abs (Figu e 2A). The i s h ee abs p o- ide in o ma ion abou mu a ions included in o combined, ene gy-based and e olu ion-based mu an . The checkbox, allowing use s o isualize he chosen esidues in Jsmol ap- ple , can be ound in each ow oge he wi h all da a ele- an o a gi en compu a ional app oach. The las ab con- ains he lis o all esidues in he wild- ype s uc u e. While ‘wild- ype’ ab is ac i e, he wild- ype s uc u e is isualized in Jsmol apple ins ead o he mu a ed one and he use is allowed o in oduce use -de ined mu a ions in o mul iple- poin mu an ia he ‘plus’ icon in he las column. Gene al in o ma ion. The ‘Fi eP o p o ocol design’ panel p o ides use s wi h gene al in o ma ion abou he a ge p o ein and he designs cons uc ed by Fi eP o s a egy, such as a numbe o mu a ions and es ima ed change in ee ene gy (Figu e 2C). Mu an designe . The ‘Mu an designe ’ panel allows he use o design own mul iple-poin mu an by managing mu- a ions di ided in o ene gy- and e olu ion-based subse . I all mu a ions in he subse ha e hei p edic ed ene gy al- ues assigned, a o al change in Gibbs ee ene gy is im- media ely es ima ed assuming simple addi i i y. Use s can also gene a e an amino acid sequence om he designed mul iple-poin mu an ha combines mu a ions included in o ene gy- and e olu ion-based subse s. All p epa ed de- signs can be downloaded in one .zip a chi e (Figu e 2E). EXPERIMENTAL VALIDATION The o iginal Fi eP o s a egy was expe imen ally e i- ied wi h h ee p o eins (haloalkane dehalogenase DhaA, PDB ID 4E46; ␥-hexachlo ocyclohexane dehyd ochlo i- nase LinA, PDB ID 3A76; and ib oblas g ow h ac o 2, PDB ID 4OEE) and p o ided espec i e s abiliza ion o p o eins Tm=25, 21 and 15◦C(Table1). The o iginal p o ocol was modi ied o enable ully au oma ed calcula- ion a he easonable ime, while main aining high p e- dic ion accu acy (Supplemen a y Table S6). P edic ion o eigh mul iple-poin mu an s using his modi ied p o ocol was alida ed using he da a o FRESCO (44) and iden i- ied mu a ions we e compa ed wi h ano he online p o ein s abiliza ion ool PROSS (17). Fi eP o and PROSS showed simila p edic i e powe , co ec ly iden i ying 29 and 20 po- en ially s abilizing posi ions, espec i ely (Supplemen a y Table S7). CONCLUSIONS AND OUTLOOK Fi eP o is a web se e ha p o ides use s wi h a one- s op-shop solu ion o he design o he mos able mul iple- poin mu an p o eins. In compa ison wi h he s andalone Fi eP o s a egy (16), all de aul pa ame e s and compu- a ional p o ocols we e op imized o inc ease he calcula- ion speed, while main aining he p edic ion accu acy. The designs p oduced by he Fi eP o wo k low we e expe - imen ally e i ied and hus use s can ob ain highly eli- able he mos able p o eins wi h minimal expe imen al e - o . The se e is complemen ed by an easy- o-use g aphi- cal in e ace ha allows use s o in e ac i ely analyze indi- idual mu a ions selec ed as a pa o ene gy- o e olu ion- based app oach oge he wi h he abili y o design hei own mul iple-poin mu an s on op o ou obus s a egy. Nucleic Acids Resea ch, 2017, Vol. 45, Web Se e issue W397 Figu e 2. Fi eP o ’s g aphical use in e ace showing he esul s ob ained o he haloalkane dehalogenase DhaA (PDB ID: 4e46). (A) The ‘Mu an o e iew’ panel p o ides a lis o mu a ions in oduced in o p o ein s uc u e. (B) The ‘Repo ’ panel shows he s a us o calcula ion in he indi idual s eps o he compu a ional pipeline. (C) The ‘P o ocol design’ panel p o ides gene al in o ma ion abou Fi eP o designs. (D) The JSmol ´Viewe ´ allows in e ac i e isualiza ion o he p o ein. (E) The ‘Mu an designe ’ panel enables manual adjus men o a new combined mu an . Table 1. Expe imen al alida ion o Fi eP o s a egy P o ein Ene gy-based mu a ions E olu ion-based mu a ions Tm[◦C] PDB ID 4E46 8 3 +25 3A76 4 3 +21 4OEE 4 2 +15 W398 Nucleic Acids Resea ch, 2017, Vol. 45, Web Se e issue The au oma ion o he whole p ocedu e makes he p o- cess o he design o he mos able p o eins accessible o use s wi hou any p io expe ise in bioin o ma ics since i elimina es he need o selec , ins all and e alua e ools, op- imize hei pa ame e s, and in e p e in e media e esul s. Howe e , he ene gy-based app oach o he Fi eP o s a - egy depends on he quali y o p o ided p o ein s uc u e and he e o e he p edic ion accu acy migh be comp o- mised in he case o low- esolu ion s uc u es o homology models. In he u u e, we plan o implemen new s a egies such as a design based on he analysis o co ela ed posi ions ha would con ibu e o he cons uc ion o he inal combined mu an , elimina ion o highly lexible egions and in oduc- ion o disul ide b idges. Also, we plan o equip Fi eP o wi h se e al new il e s, e.g. exclusion o he amino acids lo- ca ed in he close neighbo hoods o he ac i e si es o he ones pa icipa ing in oligome iza ion. SUPPLEMENTARY DATA Supplemen a y Da a a e a ailable a NAR Online. ACKNOWLEDGEMENTS We would like o hank he de elope s o PROSS Adi Goldenzweig and D Sa el Fleishman (Weizmann Ins i- u e o Science, Is ael) o p o iding he da a se o al- ida ion pu poses. Compu a ional esou ces we e supplied by he Minis y o Educa ion, You h and Spo s o he Czech Republic unde he P ojec s CESNET (P ojec No. LM2015042) and CERIT-Scien i ic Cloud (P ojec No. LM2015085) p o ided wi hin he p og am P ojec s o La ge Resea ch, De elopmen and Inno a ions In as uc u es. FUNDING This esea ch and open access cha ge was unded by Min- is y o Educa ion, You h and Spo s o he Czech Re- public om he Na ional P og amme o Sus ainabili y II [LQ1605, LO1214]; Eu opean Regional De elopmen Fund [ELIXIR-CZ LM2015047]; G an Academy o he Czech Republic [16-06096S]; B no Uni e si y Technology [FIT-S- 17-3964]. Con lic o in e es s a emen . None decla ed. REFERENCES 1. Moda es,H.P., Mo ad,M.R. and Sana i-Nezhad,A. (2016) P o ein he mos abili y enginee ing. RSC Ad .,6, 115252–115270. 2. Fe djani,S., Ioni a,M., Roy,B., Dion,M., Djeghaba,Z., Rabille ,C. and Tellie ,C. (2011) Co ela ion be ween he mos abili y and s abili y o glycosidases in ionic liquid. Bio echnol. Le .,33, 1215–1219. 3. Wijma,H.J., Floo ,R.J. and Janssen,D.B. (2013) S uc u e- and sequence-analysis inspi ed enginee ing o p o eins o enhanced he mos abili y. Cu . Opin. S uc . Biol.,23, 588–594. 4. Gao,D., Na asimhan,D.L., Macdonald,J., B im,R., Ko,M.C., Land y,D.W., Woods,J.H., Sunaha a,R.K. and Zhan,C.G. (2009) The mos able a ian s o cocaine es e ase o long- ime p o ec ion agains cocaine oxici y. Mol. Pha macol.,75, 318–323. 5. Polizzi,K.M., Bomma ius,A.S., B oe ing,J.M. and Chapa o-Rigge s,J.F. (2007) S abili y o bioca alys s. Cu . Opin. Chem. Biol.,11, 220–225. 6. G ay,K.A., Richa dson,T.H., K e z,K., Sho ,J.M., Ba nek,F., Knowles,R., Kan,L., Swanson,P.E. and Robe son,D.E. (2001) Rapid e olu ion o e e sible dena u a ion and ele a ed mel ing empe a u e in a mic obial haloalkane dehalogenase. Ad . Syn h. Ca al.,343, 607–617. 7. Folkman,L., S an ic,B., Sa a ,A. and Zhou,Y. (2016) EASE-MM: sequence-based p edic ion o mu a ion-induced s abili y changes wi h ea u e-based mul iple models. J. Mol. Biol.,428, 1394–1405. 8. Cap io i,E., Fa iselli,P. and Casadio,R. (2005) I-Mu an 2.0: p edic ing s abili y changes upon mu a ion om he p o ein sequence o s uc u e. Nucleic Acids Res., W306–W310. 9. Pi es,D.E., Asche ,D.B. and Blundell,T.L. (2014) mCSM: p edic ing he e ec s o mu a ions in p o eins using g aph-based signa u es. Bioin o ma ics,30, 335–342. 10. Kellogg,E.H., Lea e -Fay,A. and Bake ,D. (2011) Role o con o ma ional sampling in compu ing mu a ion-induced changes in p o ein s uc u e and s abili y. P o eins,79, 830–838. 11. Yin,S., Ding,F. and Dokholyan,N.V. (2007) Modeling backbone lexibili y imp o es p o ein s abili y es ima ion. S uc u e,15, 1567–1576. 12. Dehouck,Y., G os ils,A., Folch,B., Gilis,D., Bogae s,P. and Rooman,M. (2009) Fas and accu a e p edic ions o p o ein s abili y changes upon mu a ions using s a is ical po en ials and neu al ne wo ks: PoPMuSiC-2.0. Bioin o ma ics,25, 2537–2543. 13. Gue ois,R., Nielsen,J.E. and Se ano,L. (2002) P edic ing changes in he s abili y o p o eins and p o ein complexes: a s udy o mo e han 1000 mu a ions. J. Mol. Biol.,320, 369–387. 14. Gumulya,Y. and Ree z,M.T. (2011) Enhancing he he mal obus ness o an enzyme by di ec ed e olu ion: leas a o able s a ing poin s and in e io mu an s can map supe io e olu iona y pa hways. ChemBioChem,12, 2502–2510. 15. Bomma ius,A.S. and Paye,M.F. (2013) S abilizing bioca alys s. Chem. Soc. Re .,42, 6534–6565. 16. Bedna ,D., Bee ens,K., Sebes o a,E., Bendl,J., Kha e,S., Chaloupko a,R., P okop,Z., B ezo sky,J., Bake ,D. and Dambo sky,J. (2015) Fi eP o : ene gy- and e olu ion-based compu a ional design o he mos able mul iple-poin mu an s. PLoS Compu . Biol.,11, e1004556. 17. Goldenzweig,A., Goldsmi h,M., Hill,S.E., Ge man,O., Lau ino,P., Ashani,Y., Dym,O., Unge ,T., Albeck,S., P ilusky,J. e al. (2016) Au oma ed s uc u e- and sequence-based design o p o eins o high bac e ial exp ession and s abili y. Mol. Cell,63, 337–346. 18. Camacho,C., Coulou is,G., A agyan,V., Ma,N., Papadopoulos,J., Beale ,K. and Madden,T.L. (2009) BLAST+: a chi ec u e and applica ions. BMC Bioin o ma ics,10, 421. 19. Suzek,B.E., Wang,Y., Huang,H., McGa ey,P.B. and Wu,C.H. (2015) UniRe clus e s: a comp ehensi e and scalable al e na i e o imp o ing sequence simila i y sea ches. Bioin o ma ics,31, 926–932. 20. Edga ,R.C. (2010) Sea ch and clus e ing o de s o magni ude as e han BLAST. Bioin o ma ics,26, 2460–2461. 21. Sie e s,F., Wilm,A., Dineen,D., Gibson,T.J., Ka plus,K., Li,W., Lopez,R., McWilliam,H., Remme ,M., Soding,J. e al. (2011) Fas , scalable gene a ion o high-quali y p o ein mul iple sequence alignmen s using clus al omega. Mol. Sys . Biol.,7, 539. 22. Cap a,J.A. and Singh,M. (2007) P edic ing unc ionally impo an esidues om sequence conse a ion. Bioin o ma ics,23, 1875–1882. 23. Kass,I. and Ho o i z,A. (2002) Mapping pa hways o allos e ic communica ion in G oEL by analysis o co ela ed mu a ions. P o eins,48, 611–617. 24. Ko be ,B.T.M., Fa be ,R.M., Wolpe ,D.H. and Lapedes,A.S. (1993) Co a ia ion o mu a ions in he V3 loop o human immunode iciency i us ype 1 en elope p o ein: an in o ma ion heo e ic analysis. P oc. Na l. Acad. Sci. U.S.A.,90, 7176–7180. 25. Lee,B.C. and Kim,D. (2009) A new me hod o e ealing co ela ed mu a ions unde he s uc u al and unc ional cons ain s in p o eins. Bioin o ma ics,25, 2506–2513. 26. Weig ,M., Whi e,R.A., Szu man ,H., Hoch,J.A. and Hwa,T. (2008) Iden i ica ion o di ec esidue con ac s in p o ein–p o ein in e ac ion by message passing. P oc. Na l. Acad. Sci. U.S.A.,106, 67–72. 27. Lockless,S.W. and Rangana han,R. (1999) E olu iona ily conse ed pa hways o ene ge ic connec i i y in p o ein amilies. Science,286, 295–299. 28. Dekke ,J.P., Fodo ,A., Ald ich,R.W. and Yellen,G. (2004) A pe u ba ion-based me hod o calcula ing explici likelihood o Nucleic Acids Resea ch, 2017, Vol. 45, Web Se e issue W399 e olu iona y co- a iance in mul iple sequence alignmen s. Bioin o ma ics,20, 1565–1572. 29. Valencia,A. (2003) Mul iple sequence alignmen s as ools o p o ein s uc u e and unc ion p edic ion. Compa . Func . Genomics,4, 424–427. 30. Benne ,S.A. and Ge lo ,D. (1991) Pa e ns o di e gence in homologous p o eins as indica o s o seconda y and e ia y s uc u e: a p edic ion o he s uc u e o he ca aly ic domain o p o ein kinases. Ad . Enzyme Regul.,31, 121–181. 31. B enne ,S. (1988) The molecula e olu ion o genes and p o eins: a ale o wo se ines. Na u e,334, 528–530. 32. Coope man,B.S., Bayko ,A.A. and Lah i,R. (1992) E olu iona y conse a ion o he ac i e si e o soluble ino ganic py ophospha ase. T ends Biochem. Sci.,17, 262–266. 33. Howell,N. (1989) E olu iona y conse a ion o p o ein egions in he p o onmo i e cy och ome b and hei possible oles in edox ca alysis. J. Mol. E ol.,29, 157–169. 34. Gobel,U., Sande ,C., Schneide ,R. and Valencia,A. (1994) Co ela ed mu a ions and esidue con ac s in p o eins. P o eins,18, 309–317. 35. Nehe ,E. (1994) How equen a e co ela ed changes in amilies o p o ein sequences? P oc. Na l. Acad. Sci. U.S.A.,91, 98–102. 36. Taylo ,W.R. and Ha ick,K. (1994) Compensa ing changes in p o ein mul iple sequence alignmen s. P o ein Eng.,7, 341–348. 37. Ba a,K.A., G omiha,M.M., Uedai a,H., Ki ajima,K. and Sa ai,A. (2004) P oThe m, e sion 4.0: he modynamic da abase o p o eins and mu an s. Nucleic Acids Res.,32, D120–D121. 38. Pucci,F., Bou geas,R. and Rooman,M. (2016) P edic ing p o ein he mal s abili y changes upon poin mu a ions using s a is ical po en ials: in oducing HoTMuSiC. Scien i ic Rep.,6, 23257. 39. Amin,N., Liu,A.D., Rame ,S., Aehle,W., Meije ,D., Me in,M., Wong,S., Gual e i,P. and Schellenbe ge ,V. (2004) Cons uc ion o s abilized p o eins by combina o ial consensus mu agenesis. P o ein Eng. Des. Selec .,17, 787–793. 40. Lehmann,M., Loch,C., Middendo ,A., S ude ,D., Lassen,S.F., Pasamon es,L., an Loon,A.P. and Wyss,M. (2002) The consensus concep o he mos abili y enginee ing o p o eins: u he p oo o concep . P o ein Eng.,15, 403–411. 41. Pey,A.L., Rod iguez-La ea,D., Bomke,S., Damme s,S., Godoy-Ruiz,R., Ga cia-Mi a,M.M. and Sanchez-Ruiz,J.M. (2008) Enginee ing p o eins wi h unable he modynamic and kine ic s abili ies. P o eins,71, 165–174. 42. Sulli an,B.J., Nguyen,T., Du ani,V., Ma hu ,D., Rojas,S., Thomas,M., Syu,T. and Maglie y,T.J. (2012) S abilizing p o eins om sequence s a is ics: he in e play o conse a ion and co ela ion in iosephospha e isome ase s abili y. J. Mol. Biol.,420, 384–399. 43. Bendl,J., S ou ac,J., Sebes o a,E., Va a,O., Musil,M., B ezo sky,J. and Dambo sky,J. (2016) Ho Spo wiza d 2.0: au oma ed design o si e-speci ic mu a ions and sma lib a ies in p o ein enginee ing. Nucleic Acids Res.,44, W479–W487. 44. Floo ,R.J.1, Wijma,H.J., Colpa,D.I., Ramos-Sil a,A., Jekel,P.A., Szyma´ nski,W., Fe inga,B.L., Ma ink,S.J. and Janssen,D.B. (2014) Compu a ional lib a y design o inc easing haloalkane dehalogenase s abili y. ChemBioChem,15, 1660–1672.