AutoMap is a high performance homozygosity mapping tool using next-generation sequencing data

ARTICLE Au oMap is a high pe o mance homozygosi y mapping ool using nex -gene a ion sequencing da a Ma hieu Quinodoz1,2,3, Vi ginie G. Pe e 1,2,3,4, Nicola Bedoni5, Bé yl Roye Be and5, Ka a ina Cisa o a5, A ash Salmaninejad 6, Neda Sepahi7, Raquel Rod igues8, Meh an Pi an7,9, Majid Moja ad6, Ali eza Pasda 6,10, Ali Ghanba i Asad 7, Ana Be a Sousa8,11, Luisa Cou inho San os 12, And ea Supe i-Fu ga 5& Ca lo Ri ol a 1,2,3✉ Homozygosi y mapping is a powe ul me hod o iden i ying mu a ions in pa ien s wi h ecessi e condi ions, especially in consanguineous amilies o isola ed popula ions. His o i- cally, i has been used in conjunc ion wi h geno ypes om highly polymo phic ma ke s, such as DNA mic osa elli es o common SNPs. T adi ional so wa e pe o ms a he poo ly wi h da a om Whole Exome Sequencing (WES) and Whole Genome Sequencing (WGS), which a e now ex ensi ely used in medical gene ics. We de elop Au oMap, a ool ha is bo h web- based o downloadable, o allow pe o ming homozygosi y mapping di ec ly on VCF (Va ian Call Fo ma ) calls om WES o WGS p ojec s. Following a aining s ep on WES da a om 26 consanguineous amilies and a alida ion p ocedu e on a ma ched coho , ou me hod shows highe o e all pe o mances when compa ed wi h eigh exis ing ools. Mos impo an ly, when es ed on eal cases wi h nega i e molecula diagnosis om an in e nal se , Au oMap de ec s h ee gene-disease and mul iple a ian -disease associa ions ha we e p e iously un ecognized, p ojec ing clea benefi s o bo h molecula diagnosis and esea ch ac i i ies in medical gene ics. h ps://doi.o g/10.1038/s41467-020-20584-4 OPEN 1Ins i u e o Molecula and Clinical Oph halmology Basel (IOB), Basel, Swi ze land. 2Depa men o Oph halmology, Uni e si y o Basel, Basel, Swi ze land. 3Depa men o Gene ics and Genome Biology, Uni e si y o Leices e , Leices e , UK. 4Ins i u e o Expe imen al Pa hology, Lausanne Uni e si y Hospi al (CHUV), Lausanne, Swi ze land. 5Se ice o Medical Gene ics, Lausanne Uni e si y Hospi al (CHUV), Lausanne, Swi ze land. 6Depa men o Medical Gene ics, Facul y o Medicine, Mashhad Uni e si y o Medical Sciences, Mashhad, I an. 7Noncommunicable Diseases Resea ch Cen e , Fasa Uni e si y o Sciences, Fasa, I an. 8Depa men o Medical Gene ics, Hospi al San a Ma ia, Cen o Hospi ala Uni e si á io Lisboa No e (CHULN), Lisbon Academic Medical Cen e (CAML), Lisbon, Po ugal. 9Bioin o ma ics and Compu a ional Biology Resea ch Cen e , Shi az Uni e si y o Medical Sciences, Shi az, I an. 10 Di ision o Applied Medicine, Medical School, Uni e si y o Abe deen, Abe deen, UK. 11 Medical Facul y, Lisbon Uni e si y, Lisbon, Po ugal. 12 Ins i u o de O almologia D Gama Pin o, Lisbon, Po ugal. ✉email: [email p o ec ed] NATURE COMMUNICATIONS | (2021) 12:518 | h ps://doi.o g/10.1038/s41467-020-20584-4 | www.na u e.com/na u ecommunica ions 1 1234567890():,; Homozygosi y mapping (HM), also called au ozygosi y mapping, is a echnique aimed a de ec ing and sco ing he p esence o consecu i e homozygous geno ypes, o “ uns o homozygosi y”(ROHs) in a pe son’s genome. ROHs esul om he co-inhe i ance o po ions o DNA ha a e p e- alen in a gi en popula ion, and no mally ange om a ew o hund eds o megabases (Mb), depending on he e hnic g oup conside ed1,2. In he o sp ing o consanguineous unions, ROHs co espond in la ge pa o he gene ic ma e ial ha is co- inhe i ed om ances o s who a e common o bo h pa en s3. Fo ins ance, child en o pa en s who a e fi s cousins ha e ROHs o DNA egions ha we e in he e ozygosis in hei g ea -g and- pa en s, and we e inhe i ed om bo h hei ma e nal and pa e nal sides. In pa ien s wi h ecessi e condi ions, especially i belonging o a consanguineous pedig ee, ROHs e y o en encompass he mu a ion ha is esponsible o he disease4, also o igina ing om a heal hy he e ozygous common ances o , and he e o e a e excellen p oxy ma ke s o he mu a ion i sel . In addi ion, HM can also be used o de ec la ge he e ozygous dele ions esul ing in hemizygous (and appa en ly homozygous) geno ypes, as well as o cases o unipa en al disomy5,6. Fo hese easons, HM has been used o decades in medical gene ics as a ool o iden i y egions o he genome o be p io i ized o a ge ed mu a ional sc eens. In he pas , HM p ocedu es elied on geno ypes om polymo phic ma ke s ( ypically mic osa elli es) o de e mine whe he pa ien s had consecu i e homozygous calls s. heal hy ela i es. Since he beginning o his cen u y, he comme ciali- za ion o high-densi y mic oa ays o single-nucleo ide poly- mo phisms (SNPs) has allowed he use o he SNP geno ypes o he same pu pose, by in e oga ing usually non-coding a ian s wi h an ele a ed deg ee o he e ozygosi y in he gene al popula- ion. Se e al so wa e we e hen de eloped, such as PLINK7, Homozygosi yMappe 8, o GERMLINE9. In i ue o he e y high accu acy o hei calls (>99.8%)10, SNP mic oa ays ou pu s a e highly eliable o iden i y ROHs. Howe e , hey do no p o- ide any in o ma ion on a pa ien ’s mu a ions, since hese la e DNA a ian s a e usually a e and he e o e a e no included in such a ays. Con e sely, in o ma ion om whole-exome sequencing (WES), which is la gely used in con empo a y medical gene ics, allows he disco e y o any ype o DNA a ian s, including equen geno ypes and a e mu a ions alike. The e o e, a leas in p inciple, WES can be used as a single echnology o bo h HM and mu a ion de ec ion. Howe e , HM algo i hms de eloped o a ay echnologies end o deli e sub-op imal esul s when applied o WES, since hey a e no adap ed o handle he in insic noise (2.53–30.60%)10 ha is ypical o his sequencing me hod11–13. In pa icula , ROHs de ec ed om WES da a wi h his so wa e a e smalle in size and p oduce lowe cumula i e au ozygosi y alues, o en esul ing in he end in alsely-nega i e ou pu 14–16. Hence, many in es iga o s s ill make use o wo sepa a e ools: SNP mic oa ays o de e mine ROHs, and WES o iden i y possible a e mu a ions, wi h an impo an was e o ime and esou ces17–21. To ci cum en his p oblem, in his wo k we p esen Au oMap (Au ozygosi y Mappe ), a so wa e ha can p o ide e y eliable HM esul s di ec ly om s anda d WES ou pu s and is applicable o WGS (whole-genome sequencing) as an addi ional ea u e, and compa e i wi h o he ools, including some ha we e specifically de eloped o pe o m HM on WES da a (BCFTools16, FILTUS22, H3M223, HOMWES15, Sa yVc Homozygosi y, and Sa yHo- mozygosi y24). Each o he ools used in his compa ison ha e hei own s ongpoin s and limi a ions, such as he p esence o a g aphical use in e ace (GUI), a c oss-pla o m ope a i e sys em, a use - iendly ou pu ype, e c25,26. Au oMap is accessible bo h ia a web in e ace ( ee o cha ge), o allow low- h oughpu o occasional use o exome da a, and as a command-line ool, o la ge-scale genomic p ojec s, o WGS, and o allow in eg a ion in o ou ine analy ical pipelines. Resul s and discussion Au oMap akes as an inpu a ian call o ma (VCF) files (Fig. 1a), i.e., s anda d esul files om a a ie y o commonly- used so wa e o analyzing WES and WGS sequences ( a ian calle s), con aining he lis o DNA a ian s de ec ed in a gi en sample wi h espec o he human e e ence genome. VCF files ha e he ad an age o being o smalle size compa ed o p ima y mapping files (BAM files), con ain mos o he in o ma ion ha can be used o HM pu poses and, impo an ly, a e gene ally mo e a ailable o he end-use s, such as biologis s o physicians. O no e, single-sample VCFs a e accep ed by bo h web and s andalone e sions o Au oMap, while mul isample VCFs a e only accep ed by he s andalone e sion. Once he file is uploaded ia he web in e ace (h ps://au omap.iob.ch/) he use simply launches he analysis, which in he end p oduces an ou pu such as he one indica ed in Fig. 1b, c. This includes a PDF file showing he g aphical ep esen a ion o au ozygous egions along he au osomes, as well as a ex file epo ing he same in o ma ion as nume ical alues, wi h he posi ions o he de ec ed ROHs, hei size, numbe o a ian s, and pe cen age o homozygosi y. Au oMap can also include he X ch omosome, i eques ed by he use , o bo h emales and males. In emales, his ch omosome is ea ed as an au osome, whe eas in males all hemizygous calls a e conside ed as homozygous. In addi ion, i is possible o p o ide a lis o genes (o a “gene panel”, e.g., a lis o genes linked o a gi en condi ion) as a ex file, o immedia ely ecognize whe he hey a e o no wi hin a gi en ROH (Fig. 1c). The same p ocedu e can also be pe o med on a local compu e , by di ec ly down- loading he sou ce files (h ps://gi hub.com/mquinodo/ Au oMap). In i s s andalone e sion, Au oMap does no need compiling bu equi es he ins alla ion o some addi ional so - wa e (BCFTools, 1.9 o la e ; BEDTools, 2.25.0 o la e ; Pe l, 5.22.0 o la e ; R, 3.2.0 o la e ). As men ioned, one o he key poin s o he p ope de ec- ion o ROH om WES da a is he a ge ed emo al o alse posi i e a ian calls. These calls no only ac as a noise wi h espec o ue signals, bu also ac i ely lead o an a ificial agmen a ion o ROHs (when a he e ozygous alse posi i e call is p esen in a uly au ozygous egion) o e en o he comple e miss o a ROH ( o egions wi h a low numbe o calls). To his end, as a fi s s ep, all a ian s om he VCF file a e ca e ully assessed wi h c i e ia elying on co e age and al e na i e eads coun . Mo e p ecisely, hey a e e alua ed acco ding o hei sequencing dep h, he a io be ween eads aligned o he e e ence sequence s. al e na i e alleles ( o he e ozygous calls), and o hei loca ion in epea ed egions o he genome (Fig. 1a and Supplemen a y Table 1). Va ian s ha do no sa is y hese s ingen c i e ia a e selec i ely and indi idually elimina ed om u he analyses. Impo an ly, all such c i e ia we e e ified no o be specific o any gi en a ian calling so wa e, wi h he aim o enabling analyses o VCF files p oduced by a ious p og ams. A e his fil e ing s ep a he a ian le el, ROHs a e iden ified by a sliding-window app oach, associa ed wi h ou unable pa ame e s: he size o he window, he minimal numbe o homozygous a ian s in he window, he maximal gap be ween wo consecu i e a ian s in one ROH, and he maximal size o ex ension om he bounda ies o s e ches o de ec ed ROHs (Fig. 1a and Supplemen a y Table 1). Following hei iden ifica- ion, ROHs a e hen selec ed only i hey each a minimal alue in ARTICLE NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-020-20584-4 2NATURE COMMUNICATIONS | (2021) 12:518 | h ps://doi.o g/10.1038/s41467-020-20584-4 | www.na u e.c om/na u ecommunica ions e ms o size, numbe o a ian s and pe cen age o homozygous geno ypes in he egion (Fig. 1a and Supplemen a y Table 1). To e alua e he pe o mances o Au oMap, we ga he ed da a om 52 amilies wi h ecessi e he edi a y blindness, o which we also had geno ypes om bo h WES and SNP a ays, and selec ed one pa ien pe amily. These indi iduals we e all om consanguineous pedig ees li ing in Po ugal o in I an, i.e., in coun ies wi h a low and an ele a ed deg ee o consanguini y a he le el o he gene al popula ion, espec i ely27,28. Mo e p e- cisely, hey all had a median cumula i e ROH size o 234.8 Mb (de ec ed by PLINK on SNP a ay da a), anging be ween 90.6 and 819.5 Mb. We hen spli hese indi iduals in o wo ma ched se s o 26 pe sons each, hus defining a aining coho and a ma ched alida ion coho (Supplemen a y Da a 1). As a e e - ence o ue posi i e alues, we adop ed he me hod desc ibed by Kanche a e al.15, consis ing in he use o he PLINK so wa e7 applied wi h de aul pa ame e s on da a om SNP a ays. We op imized he pa ame e s o Au oMap on he aining coho (Supplemen a y Table 1 and Supplemen a y Fig. 1A, B, see “Me hods”sec ion), and hen used he op imized alues o ana- lyze he alida ion coho . We pe o med s abili y analyses by a ying each pa ame e indi idually, which esul ed in small a ia ion o he pe o mances, indica i e o he obus ness o he me hod (Supplemen a y Fig. 1C). In addi ion, he a e age pe - o mance was no significan ly di e en be ween he wo coho s, demons a ing no o e -fi ing o he pa ame e s on he aining da a (Supplemen a y Fig. 2A). Mo e p ecisely, specifici y in he aining and alida ion se s had alues o 77.5% and 79.6%, espec i ely (p=0.35, unpai ed - es ), while sensi i i y had alues o 90.5% and 92.4%, espec i ely (p=0.18). We also compa ed he e ec o h ee mains eam a ian calle s ha could be adop ed by Au oMap end use s (GATK- Haplo ype Calle , Sam ools-mpileup and S elka) on he pe o - mance o ou ool (based on da a om he aining se ). Au o- Map’s sensi i i y was sligh ly lowe when S elka was used (−1.4% compa ed o Haplo ype Calle ), whe eas specifici y appea ed o be sligh ly lowe when Sam ools was used, (−4.4% compa ed o Haplo ype Calle , Supplemen a y Fig. 2B). These di e ences we e all below 5%, he e o e indica ing ha ou ool is o e all insensi i e o he choice ope a ed by a po en ial use wi h espec o a gi en calling so wa e. Nex , we compa ed Au oMap wi h eigh p e iously-published ools, namely: PLINK applied on exome da a7, Homo- zygosi yMappe o WES29, HOMWES15, BCFTools/RoH16, FILTUS22,H 3M223, Sa yHomozygosi y, and Sa yVc Homo- zygosi y24 on da a om he alida ion coho , composed o 26 indi iduals wi h a ious le els o au ozygosi y (Supplemen a y Fig. 3). All o hem we e based on a hidden Ma ko model wi h he excep ion o Homozygosi yMappe , PLINK, and HOMWES, which we e de eloped by applying a sliding window me hod (as i is he case o Au oMap as well). O no e, H3M2and Sa yHo- mozygosi y canno p ocess VCF files bu need a BAM file (sequence alignmen da a), Homozygosi yMappe is only a ail- able ia a web in e ace, and FILTUS can only be que ied ia a Fig. 1 Au oMap wo kflow and example o ou pu . a Wo kflow ollowed by Au oMap wi h de aul se ings, b,cexample o g aphical and ex ou pu o pa ien NSI-326, wi h he ollowing pa ame e s: DP =8, pe cal low =0.25, pe cal high =0.75, binomial =0.000001, maxgap =10, window =7, window h es =5, minsize =2, min a =25, minpe c =88, ch X =No, and ex end =1. Blue egions ep esen de ec ed ROHs. NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-020-20584-4 ARTICLE NATURE COMMUNICATIONS | (2021) 12:518 | h ps://doi.o g/10.1038/s41467-020-20584-4 | www.na u e.com/na u ecommunica ions 3 g aphical use in e ace (GUI), no allowing au oma ed p oces- sing o mul iple files. The use Sa yHomozygosi y and Sa - yVc Homozygosi y equi es he downloading o p e-p epa ed linkage da a om 1000 Genomes p ojec o a join c om a ew hund ed WGS. We did no include o he so wa e packages such as Agile-Geno ype , Agile-Va ian Mappe 14, and HomSI30 since hey do no p o ide a genome-wide esul file, which could be used as sou ce o da a o compa ison, bu only a g aphical ou pu . We used de aul pa ame e s o all hese ools (i.e., he pa ame e s op imized by hei espec i e au ho s o analyze WES da a), excep o PLINK, since his so wa e was fi s de eloped o SNP a ay da a and he e o e had o be e-pa ame ized o allow i s use on WES geno ypes (see “Me hods”sec ion). Au oMap and Sa yVc Homozygosi y had he bes sensi i i y/ specifici y combina ion and he highes F-sco e when compa ed o o he so wa e (p< 3.0 × 10−7, pai ed - es ) (Fig. 2a, b). Howe e , Au oMap had a significan ly highe sensi i i y han Sa yVc Homozygosi y (p< 2.9 × 10−6) and lowe specifici y (p< 2.8 × 10−6). In addi ion, Au oMap (on WES) was he ool ha displayed he closes alues, in e ms o ROHs size and numbe , o he e e ence (i.e., PLINK on SNP a ay) (Fig. 3). This indica ed no o e y low ROH agmen a ion, especially in compa ison o o he ools, which p oduced o ins ance ei he a high numbe o small and agmen ed ROHs (H3M2, FILTUS, and PLINK) o de ec ed only la ge ROHs (Sa yVc Homozygosi y). Two ep esen a i e examples o his e ec , o e en i e cho omosomes, a e p esen ed in Supplemen a y Fig. 4. Since small ROHs a e mo e di ficul o de ec and indeed could penalize he global pe o mance o some o hese ools, we epea ed he same analysis by conside ing only ROHs la ge han 5 Mb, being awa e ha such egions a e mo e likely o ha bo causa i e a ian s o ecessi e diseases4. As expec ed, e e y ool displayed inc eased pe o mances, due o he a ificial clea ing o Fig. 2 Pe o mance o Au oMap and o he ools on da a om he alida ion se . a,bPe o mance o Au oMap o ROHs la ge han 1 Mb (Megabase): aspecifici y and sensi i i y; bF-sco e. c,dSame analyses as in a,b, bu limi ed o ROHs wi h sizes o 5 Mb o highe (mos likely o con ain causa i e ecessi e mu a ions in medical gene ics p ac ice, acco ding o published li e a u e). E o ba s ep esen s anda d de ia ions o he mean. Fo boxplo s (b,d), he middle band indica es he median, boxes ep esen he fi s and he hi d qua iles, and whiske s indica e he la ges obse a ion smalle han o equal o he fi s qua ile −1.5 x IQR ( he in e qua ile ange) and he smalles obse a ion g ea e han o equal o o hi d qua ile +1.5 x IQR. N= ~2.7 million DNA a ian s pe ool and pe es . Sou ce da a a e p o ided in he Sou ce Da a File. ARTICLE NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-020-20584-4 4NATURE COMMUNICATIONS | (2021) 12:518 | h ps://doi.o g/10.1038/s41467-020-20584-4 | www.na u e.c om/na u ecommunica ions smalle ROHs, and Au oMap, Sa yHomozygosi y and Sa - yVc Homozygosi y had he bes F-sco es (p< 2.0 × 10−4, Fig. 2c, d). Au oMap had a significan highe sensi i i y han Sa yVc - Homozygosi y (p=2.9 × 10−6) bu lowe han Sa yHomo- zygosi y (p< 6.9 × 10−9). Conce ning specifici y, he esul s we e e e sed, wi h Au oMap ha ing a significan ly highe alue han Sa yHomozygosi y (p< 3.0 × 10−3) bu a lowe sco e wi h espec o Sa yVc Homozygosi y (p< 2.5 × 10−5). In summa y, Au oMap had he bes pe o mances o e all when es ed on ou alida ion coho . Since his coho comp ised only indi iduals o Eu opean and Sou h Asian ances y, u he assessmen s may be needed o e alua e popula ion-specific pe o mances on da a om o he e hnic g oups. In addi ion, a possible con ounding ac o in ou analysis could be ep esen ed by he use o di e en cap u e ki s o WES. Al hough we did no assess his pa ame e a he expe imen al le el, p e ious li e a u e has shown ha conco dance a es o called a ian s ac oss di e en cap u e ki s a e e y ele a ed31, making subs an ial a iabili y o pe o mance linked o he use o di e en ki s a a he unlikely e en . Con- ce ning WGS da a, addi ional es s will be also needed o benchma k exis ing ools such as PLINK o o he a ay-specific so wa e s. Au oMap, due o he di e ence in e ms o numbe o a ian s and co e age compa ed o WES. In addi ion o an inc eased pe o mance in e ie ing ROHs om exome da a, Au oMap also displayed o he benefi s. Fo ins ance, Au oMap was he only ool ha could be un bo h ia a web applica ion ( o he a e age use ) and he command-line, o allow mo e compu a ionally-inclined use s o exploi i s ull po en ial. In addi ion, he a e age p ocessing ime displayed by Au oMap was ela i ely sho , especially in ela ionship o i s pe o mance. All ools, excep hose using BAM files as an inpu (H3M2and Sa yHomozygosi y) could analyze a s anda d VCF in 30 s o less. The as es ools we e BCFTools and PLINK, comple ing he analysis in less han 5 s, while Au oMap ook app oxima ely 20 s. A mo e de ailed compa ison o all hese ea u es is p o ided in Supplemen a y Table 2. As a final es , we used Au oMap on unexplo ed WGS and WES esea ch da a om ou labo a o y, o cases wi h unknown molecula e iology. Following expe imen al alida ion, ou ool was ins umen al in he end o iden i ying bo h coding and noncoding mu a ions, all included in o la ge ROHs, o a ew new synd omes. These included: an in e genic dele ion causing de elopmen al de ec s32, a small dele ion a ec ing splicing in he gene PISD and esul ing in he Libe a b synd ome33, and a pa ial duplica ion o he NMNAT1 gene, esponsible o a new mul isys em diso de 34. Compa ed o he o he ools analyzed, Au oMap was he only one allowing he comple e de ec ion o he au ozygous egions con aining he mu a ions in all h ee cases (Supplemen a y Table 3). Fo he NMNAT1 s udy, o example, he lack o pe o mance displayed by some o he ools was likely due o he p esence o a ew alse-posi i e he e ozygous a ian s in he ROH con aining he pa hogenic duplica ion, which we e indeed ecognized as such—and hence disca ded—by Au oMap. The same was ue o he wo o he s udies, whe e he p esence o alse-posi i e he e ozygous geno ypes was combined wi h he ela i ely sho size o ROHs and/o he p esence o a limi ed numbe o iden ifiable geno ypes. In addi ion, Au oMap enabled he de ec ion o new a ian s in a numbe o genes ha we e al eady associa ed wi h Mendelian condi ions35–37. In conclusion, Au oMap is a eliable ool ha can p edic ROHs wi h high specifici y and sensi i i y, in less han a minu e, e en om VCF files de i ed om noisy exome sequencing da a. I is a ailable bo h ia a web-based in e ace, o a quick analysis, as well as a command-line package, allowing la ge-scale and ou ine analyses. Me hods Pa ien s and DNA. This s udy adhe ed o he o he ene s o he Decla a ion o Helsinki and was app o ed by he Ins i u ional Re iew Boa ds o ou espec i e ins i u ions: he E hikkommission No dewes - and Zen alschweiz (2019-01660), he Ins i u ional Re iew Boa ds o Mashhad Uni e si y o Medical Sciences (961015), he e hics commission o he Oph halmic Hospi al “D Gama Pin o”in Lisbon (16.05.20) and he Noncommunicable Diseases Resea ch Cen e o Fasa Uni e si y o medical sciences (IR.FUMS.REC.1396.211). W i en in o med con- sen o ms we e signed by all subjec s, ec ui ed a he Oph halmic Hospi al “D Gama Pin o”in Lisbon, and a he Fasa and Mashhad Uni e si ies o Medical Science in I an. Pa icipan s had ei he epo ed consanguini y o had ~100 Mb o mo e o cumula i e ROHs as de e mined by PLINK applied on a ay da a (Sup- plemen a y Da a 1). Genomic DNA was ex ac ed om pe iphe al blood leuko- cy es. Mo e p ecisely, o Po uguese pa ien s DNA ex ac ion was pe o med by using he EZ1 DNA blood ki and EZ1 DNA bu y coa ca d (Qiagen), acco ding o he manu ac u e ’s ins uc ions, choosing an elu ion olume o 200 μl. Fo I anian pa ien s, DNA was ex ac ed om blood using RPN8512 Nucleon BACC3 DNA Ex ac ion Ki (Illus a). A ay geno yping. DNA o s udied indi iduals we e geno yped a he iGE3 Pla o m o he Uni e si y o Gene a, Swi ze land, using Illumina Infinium a ays (San Diego, USA; GSAMD-24 2.0, GSA-24 2.0, Co eExome-24 1.1, and Co eExome-24 1.2). Geno ypes alues we e ob ained wi h GenomeS udio (Illumina). Exome sequencing. Exome cap u e and lib a y p epa a ion was pe o med using he Su eSelec Human All Exon 6 ki (Agilen , San aCla a, USA) and HiSeq Rapid PE Clus e Ki 2 (Illumina, San Diego, USA) wi h 2 μg genomic DNA. Lib a ies we e sequenced on a HiSeq 2500 o a No aSeq 6000 ins umen s (Illumina) (Supplemen a y Da a 1). Raw eads we e mapped o he human genome e e ence sequence (build hg19) wi h he No oalign so wa e (V3.08.00, No oc a Tech- nologies, Selango , Malaysia). Duplica e eads we e hen emo ed using Pica d ( . 2.14.0-SNAPSHOT). Base quali y sco e ecalib a ion was pe o med and a ian calling was done wi h Haplo ypeCalle (GATK, .4.0.3.0). Au oMap equi emen s. Au oMap is composed o Bash, Pe l and R sc ip s. I equi es BCFTools (≥ 1.9), BEDTools (≥ 2.25.0), Pe l (≥ 5.22.0), and R (≥ 3.2.0). The ollowing e sions we e used o all analysis: BCFTools ( 1.9-78-gb7e4ba9), BEDTools ( 2.25.0), Pe l ( 5.22.0), Bash (4.3.48(1)- elease) and R ( 3.5.1). To be p ocessed by Au oMap, a VCF file mus con ain he GT (geno ype) and AD (allelic dep hs o he e and al alleles) o DP4 (numbe o high-quali y e - o wa d bases, e - e e se, al - o wa d, and al - e e se bases) fields. F H V S A 3 P WB 0 4 8 12 0 40 80 120 160 200 A e age numbe o ROHs Median o ROH size a e ages [Mb] Au oMap BCFTools Fil us H3M2 HOMWES A B F 3 W Homozygosi yMappe PLINK Sa yHomozygosi y Sa yVc Homozygosi y SNP−a ay PLINK H P S V * Fig. 3 A e age numbe o ROH de ec ed pe sample (N=26) s. he median o he size a e ages o ROHs pe sample (ROHs la ge han 1Mb).The as e isk indica es he e e ence alue used o compa ison, i.e., da a om SNP a ays analyzed wi h PLINK. E o ba s ep esen s anda d de ia ions o he mean. Sou ce da a a e p o ided in he Sou ce Da a File. NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-020-20584-4 ARTICLE NATURE COMMUNICATIONS | (2021) 12:518 | h ps://doi.o g/10.1038/s41467-020-20584-4 | www.na u e.com/na u ecommunica ions 5 Au oMap algo i hm. The fi s s ep o he algo i hm emo es a ian s om he VCF file ha a e loca ed in epea egions, as epo ed by he UCSC genome b owse (h ps://genome.ucsc.edu/cgi-bin/hgTables). Va ian s a e hen fil e ed by quali y, based on he pe cen age o al e na i e eads (de aul op ions: –minpe cal 0.25 and –maxpe cal 0.75), on a binomial es o al e na i e and e e ence ead coun s o he e ozygous a ian s (de aul op ion: –binomial 0.000001), and on dep h (de aul op ion: –DP 8). A his poin , he analysis is s opped i he e a e less han 10,000 a ian s su i ing such p ocedu es. ROHs a e subsequen ly de ec ed by a sliding window wi h a fixed size and a h eshold based on he numbe o homozygous a ian s (de aul op ions: –window 7 a ian s and –window h es 5). A e ha , de ec ed ROHs a e immed a hei ends o emo e he e ozygous a ian s and a e ex ended a bo h ends (de aul op ion: –ex end 1 [Mb]). ROHs con aining egions wi hou a ian s o a s e ch la ge han a gi en h eshold (de aul op ion: –maxgap 10 [Mb]) a e spli in o wo di e en ROHs by excluding he egion wi h no a ian s. Finally, ROHs a e fil e ed o ha e a minimal size, a minimal numbe o a ian s and a minimal pe cen age o homozygous a ian s (de aul op ions: –minsize 1 [Mb], –min a 25, and –minpe c 88 [%]). P obabili ies o binomial dis ibu ion. P obabili ies o binomial dis ibu ions we e calcula ed wi h he Pe l sc ip w i en by T.J. Finney (h ps://www.halo ype. com/RKM/figu es/TJF/binomial. x ). O e lap o ROHs. O e lap o ROHs we e ob ained wi h bed ools and he ol- lowing command: bed ools in e sec -a a.bed -b b.bed Pa ame e s used o de ec ROHs om a ay da a. PLINK was used wi h de aul pa ame e s, excep o a ian s wi h mino allele coun o 2 pe ba ch, o fil e ing agains e y a e a ian s, o en ep esen ing alse-posi i e38,39: plink –bfile bfile –homozyg –ou ou –homozyg-window-he 1 –homozyg-densi y 50 –homozyg-gap 1000 –homozyg-window-missing 5 –homozyg-window-snp 50 – homozyg-snp 100 –homozyg-window- h eshold 0.05 –homozyg-kb 1000 –mac 2 De ec ion o ROHs om exome da a. All a ailable ools we e used wi h de aul pa ame e s o ea ing inpu files, op imized by hei de elope s, excep o PLINK. Fo his ool, we ook he pa ame e s defined by Kanche a e al. 15, o op imize sensi i i y, and e- es ed he mos impo an pa ame e , i.e., he numbe o allowed he e ozygous SNP pe ROH. This pa ame e is impo an o ake in accoun he noise in exome sequencing da a. Indeed, low alues o his pa ame e esul ed in low sensi i i y. We chose o use a alue o 3 since i p oduced a highe sensi i i y wi h accep able specifici y (Supplemen a y Fig. 5). The command used was: plink –bfile bfile –homozyg –ou ou –homozyg-kb 1000 –homozyg-window-he 3 –homozyg-densi y 10000 –homozyg-gap 10000 –homozyg-window-missing 10 – homozyg-window-snp 20 –homozyg-snp 10 –homozyg-window- h eshold 0.05 Fo Sa yHomozygosi y and Sa yVc Homozygosi y, we used he p e-p epa ed linkage da a om 1000 Genomes p ojec , p o ided by he de elope s (h ps:// gi hub.com/ demolgen/Sa ySui e). Pa ame e used o a ian calle s. Sam ools: sam ools mpileup - DP,AD -ug e . as a inpu .bam|bc ools call - mO -o ou . c S elka: configu eS elkaGe mlineWo kflow.py –bam inpu .bam – e e enceFas a e . as a –exome Haplo ypeCalle : ga k –ja a-op ions “-Xmx4g”Haplo ypeCalle -R e . as a –dbsnp dbsnp. c -I $inpu .bam -O ou . c A e calling, a ian s we e fil e ed o be p esen in he egions included in he cap u e ki s, wi h a 100 bp padding on each side. Compu a ions o pe o mances. As a e e ence o ue posi i e alues, we adop ed he me hod desc ibed by Kanche a e al.15, consis ing in he use o he PLINK so wa e7applied wi h de aul pa ame e s on da a om SNP a ays. Sen- si i i y and specifici y we e hen compu ed as ollows: Sensi i i y %ðÞ¼ROHexomeall o e lapping wi h ROHa ay il e ed=To al ROHa ay il e ed; Speci ici y %ðÞ¼ROHexome il e ed o e lapping wi h ROHa ayall=To al ROHexome il e ed; whe e fil e ed ROHs a e la ge han 1 o 5 Mb and excluding gap egions such as cen ome es, elome es, sho a ms, and he e och oma in defined om he gap able in UCSC Table B owse (h ps://genome.ucsc.edu/cgi-bin/hgTables). F-sco e was also added as a measu e o accu acy, compu ed as: Fsco e ¼2sensi i i y speci ici y=sensi i i y þspeci ici yðÞ: S a is ical es s.T- es we e pe o med wi h . es unc ion in RS udio ( 1.0.153) wi h R ( 3.5.1). Fo pai ed es : . es (x, y, al e na i e =“ wo.sided”, pai ed =TRUE) Fo unpai ed es : . es (x, y, al e na i e =“ wo.sided”, pai ed =FALSE) Figu es. Figu es we e done wi h ggplo 2 ( 3.3.2) and g idEx a ( 2.3) packages in RS udio ( 1.0.153) wi h R ( 3.5.1). Repo ing summa y. Fu he in o ma ion on esea ch design is a ailable in he Na u e Resea ch Repo ing Summa y linked o his a icle. Da a a ailabili y A ay geno yping da a and exome sequencing da a canno be sha ed because o es ic ions ela ed o he p o ec ion o pe sonal da a, as pe Swiss and Eu opean law. Da a om Figs. 2and 3can be e ie ed in he Sou ce Da a file. Sou ce da a a e p o ided wi h his pape . Code a ailabili y Au oMap code is eely a ailable on Gi Hub (h ps://gi hub.com/mquinodo/Au oMap/) and was deposi ed in he zenodo.o g eposi o y (h ps://doi.o g/10.5281/ zenodo.4279125)40. Recei ed: 17 Ma ch 2020; Accep ed: 9 Decembe 2020; Re e ences 1. Ceballos, F.C., Joshi, P.K., Cla k, D.W., Ramsay, M. & Wilson, J.F. Runs o homozygosi y: windows in o popula ion his o y and ai a chi ec u e. Na Re Gene 19, 220–234 (2018). 2. Pembe on, T. J. e al. Genomic pa e ns o homozygosi y in wo ldwide human popula ions. Am. J. Hum. Gene . 91, 275–292 (2012). 3. McQuillan, R. e al. Runs o homozygosi y in Eu opean popula ions. Am. J. Hum. Gene . 83, 359–372 (2008). 4. Wakeling, M. N. e al. Homozygosi y mapping p o ides suppo ing e idence o pa hogenici y in ecessi e Mendelian disease. Gene . Med. 21, 982–986 (2019). 5. Bis, D. M. e al. Unipa en al disomy de e mined by whole-exome sequencing in a spec um o a e mo oneu on diseases and a axias. Mol. Gene . Genom. Med. 5, 280–286 (2017). 6. Yauy, K., de Leeuw, N., Yn ema, H. G., P und , R. & Gilissen, C. Accu a e de ec ion o clinically ele an unipa en al disomy om exome sequencing da a. Gene . Med. 22, 803–808 (2019). 7. Pu cell, S. e al. PLINK: a ool se o whole-genome associa ion and popula ion-based linkage analyses. Am. J. Hum. Gene . 81, 559–575 (2007). 8. Seelow, D., Schuelke, M., Hildeb and , F. & Nu nbe g, P. Homozygosi yMappe —an in e ac i e app oach o homozygosi y mapping. Nucleic Acids Res. 37, W593–W599 (2009). 9. Guse , A. e al. Whole popula ion, genome-wide mapping o hidden ela edness. Genome Res. 19, 318–326 (2009). 10. Yi, M. e al. Pe o mance compa ison o SNP de ec ion ools wi h illumina exome sequencing da a—an assessmen using bo h amily pedig ee in o ma ion and sample-ma ched SNP a ay da a. Nucleic Acids Res. 42, e101 (2014). 11. Pa el, Z. H. e al. The s uggle o find eliable esul s in exome sequencing da a: fil e ing ou Mendelian e o s. F on . Gene . 5, 16 (2014). 12. Ca son, A. R. e al. E ec i e fil e ing s a egies o imp o e da a quali y om popula ion-based whole exome sequencing s udies. BMC Bioin o m. 15, 125 (2014). 13. De Summa, S. e al. GATK ha d fil e ing: unable pa ame e s o imp o e a ian calling o nex gene a ion sequencing a ge ed gene panel da a. BMC Bioin o m. 18, 119 (2017). 14. Ca , I. M. e al. Au ozygosi y mapping wi h exome sequence da a. Hum. Mu a . 34,50–56 (2013). 15. Kanche a, D. e al. No el mu a ions in genes causing he edi a y spas ic pa aplegia and Cha co -Ma ie-Too h neu opa hy iden ified by an op imized p o ocol o homozygosi y mapping based on whole-exome sequencing. Gene . Med. 18, 600–607 (2016). 16. Na asimhan, V. e al. BCF ools/RoH: a hidden Ma ko model app oach o de ec ing au ozygosi y om nex -gene a ion sequencing da a. Bioin o ma ics 32, 1749–1751 (2016). 17. Gund, C., Powis, Z., Alca az, W., Desai, S. & Ba anano, K. Iden ifica ion o a synd ome comp ising mic ocephaly and in ellec ual disabili y bu no whi e ARTICLE NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-020-20584-4 6NATURE COMMUNICATIONS | (2021) 12:518 | h ps://doi.o g/10.1038/s41467-020-20584-4 | www.na u e.c om/na u ecommunica ions ma e disease associa ed wi h a homozygous c.676C>T p.R226W DEAF1 mu a ion. Am. J. Med. Gene . A 170A, 1330–1332 (2016). 18. Liu, L. e al. A no el hemizygous SACS mu a ion iden ified by whole exome sequencing and SNP a ay analysis in a Chinese ARSACS pa ien . J. Neu ol. Sci. 362, 111–114 (2016). 19. Ba ne , C. P. e al. Ec odac yly and le hal pulmona y acina dysplasia associa ed wi h homozygous FGFR2 mu a ions iden ified by exome sequencing. Hum. Mu a . 37, 955–963 (2016). 20. Umai , M. e al. Fi s di ec e idence o in ol emen o a homozygous loss-o - unc ion a ian in he EPS15L1 gene unde lying spli -hand/spli - oo mal o ma ion. Clin. Gene . 93, 699–702 (2018). 21. Zhang, D. e al. Homozygosi y mapping and whole exome sequencing e eal a no el ERCC8 mu a ion in a Chinese consanguineous amily wi h unique ce ebella a axia. Clin. Chim. Ac a 494,64–70 (2019). 22. Vigeland, M. D., Gjo e ud, K. S. & Selme , K. K. FILTUS: a desk op GUI o as and e ficien de ec ion o disease-causing a ian s, including a no el au ozygosi y de ec o . Bioin o ma ics 32, 1592–1594 (2016). 23. Magi, A. e al. H3M2: de ec ion o uns o homozygosi y om whole-exome sequencing da a. Bioin o ma ics 30, 2852–2859 (2014). 24. Wakeling, M., De F anco, E., Ha e sley, A. & Ella d, S. Making he mos o a ge ed sequencing: de ec ing CNVs and homozygous egions using o - a ge eads wi h Sa yCNV. Abs ac #104. P esen ed a he 67 h Annual Mee ing o he Ame ican Socie y o Human Gene ics (Oc obe 18, 2017 in O lando, Fo ida). 25. Oli ei a, J., Pe ei a, R., San os, R. & Sousa, M. Homozygosi y mapping using whole-exome sequencing: a aluable app oach o pa hogenic a ian iden ifica ion in gene ic diseases. in In e na ional Con e ence on Bioin o ma ics Models, Me hods and Algo i hms Vol. 4 210–216 (SCITEPRESS, Se ubal, 2017). 26. Oli ei a, J., Pe ei a, R., San os, R. & Sousa, M. E alua ing uns o homozygosi y in exome sequencing da a-u ili y in disease inhe i ance model selec ion and a ian fil e ing. in In e na ional Join Con e ence on Biomedical Enginee ing Sys ems and Technologies 268–288 (Sp inge , New Yo k, 2017). 27. No hnagel, M., Lu, T. T., Kayse , M. & K awczak, M. Genomic and geog aphic dis ibu ion o SNP-defined uns o homozygosi y in Eu opeans. Hum. Mol. Gene . 19, 2927–2935 (2010). 28. Saada , M., Ansa i-La i, M. & Fa hud, D. D. Consanguineous ma iage in I an. Ann. Hum. Biol. 31, 263–269 (2004). 29. Seelow, D. & Schuelke, M. Homozygosi yMappe 2012–b idging he gap be ween homozygosi y mapping and deep sequencing. Nucleic Acids Res. 40, W516–W520 (2012). 30. Go mez, Z., Baki -Gungo , B. & Sagi oglu, M. S. HomSI: a homozygous s e ch iden ifie om nex -gene a ion sequencing da a. Bioin o ma ics 30, 445–447 (2014). 31. Shigemizu, D. e al. Pe o mance compa ison o ou comme cial human whole-exome cap u e pla o ms. Sci. Rep. 5, 12742 (2015). 32. Allou, L. e al. The human eng ailed-1 (EN1) locus; wo di e en ypes o mu a ions esul ing in limb and b ain mal o ma ion pheno ypes. Am. J. Med. Gene . Pa A 179, 707–707 (2019). 33. Pe e , V. G. e al. The Libe a b synd ome, a mul isys em diso de a ec ing eye, ea , bone, and b ain de elopmen , is caused by a ounde pa hogenic a ian in he PISD gene. Gene . Med. 21, 2734–2743 (2019). 34. Bedoni, N. e al. An Alu-media ed duplica ion in NMNAT1, in ol ed in NAD biosyn hesis, causes a no el synd ome, SHILCA, a ec ing mul iple issues and o gans. Hum. Mol. Gene . 29, 2250–2260 (2020). 35. Pe e , V. G. e al. A no el missense a ian in IDH3A causes au osomal ecessi e e ini is pigmen osa. Oph halmic Gene . 40, 177–181 (2019). 36. Moye, A. R. e al. Mu a ions in ARL2BP, a p o ein equi ed o cilia y mic o ubule s uc u e, cause synd omic male in e ili y in humans and mice. PLoS Gene . 15, e1008315 (2019). 37. Rehman, A. U. e al. Explo ing he gene ic landscape o e inal diseases in No h-Wes e n Pakis an e eals a high deg ee o au ozygosi y and a p e alen ounde mu a ion in ABCA4. Genes 11, 12 (2019). 38. W igh , C. F. e al. Assessing he pa hogenici y, pene ance, and exp essi i y o pu a i e disease-causing a ian s in a popula ion se ing. Am. J. Hum. Gene . 104, 275–286 (2019). 39. Weedon, M. N. e al. Assessing he analy ical alidi y o SNP-chips o de ec ing e y a e pa hogenic a ian s: implica ions o di ec - o-consume gene ic es ing. bioRxi 696799 (2019). 40. Quinodoz, M. e al. Au oMap: High Pe o mance Homozygosi y Mapping using Nex -Gene a ion Sequencing Da a (Zenodo, 2020). Acknowledgemen s This wo k was suppo ed by he Swiss Na ional Science Founda ion (g an # 176097, o C.R.) and by he PhD Fellowships in Li e Science o he Uni e si y o Lausanne ( o M.Q.). Au ho con ibu ions M.Q. and C.R. designed he esea ch, w o e he manusc ip , analyzed all he da a, and in e p e ed he esul s. V.P., N.B., A.S., N.S., R.R., M.P., M.M., A.P., A.G.A., A.B.S., and L.C-.S. pa icipa ed o he collec ion o biological samples and DNA ex ac ion. V.P., N.B., B.R.B., K.C., and A.S.-F. pa icipa ed o he expe imen al design and p o ided commen s. Compe ing in e es s The au ho s decla e no compe ing in e es s. Addi ional in o ma ion Supplemen a y in o ma ion is a ailable o his pape a h ps://doi.o g/10.1038/s41467- 020-20584-4. Co espondence and eques s o ma e ials should be add essed o C.R. Pee e iew in o ma ion Na u e Communica ions hanks Ryan Dhindsa, Sla é Pe o ski and he o he , anonymous, e iewe o hei con ibu ion o he pee e iew o his wo k. Rep in s and pe mission in o ma ion is a ailable a h p://www.na u e.com/ ep in s Publishe ’s no e Sp inge Na u e emains neu al wi h ega d o ju isdic ional claims in published maps and ins i u ional a filia ions. Open Access This a icle is licensed unde a C ea i e Commons A ibu ion 4.0 In e na ional License, which pe mi s use, sha ing, adap a ion, dis ibu ion and ep oduc ion in any medium o o ma , as long as you gi e app op ia e c edi o he o iginal au ho (s) and he sou ce, p o ide a link o he C ea i e Commons license, and indica e i changes we e made. The images o o he hi d pa y ma e ial in his a icle a e included in he a icle’s C ea i e Commons license, unless indica ed o he wise in a c edi line o he ma e ial. I ma e ial is no included in he a icle’s C ea i e Commons license and you in ended use is no pe mi ed by s a u o y egula ion o exceeds he pe mi ed use, you will need o ob ain pe mission di ec ly om he copy igh holde . To iew a copy o his license, isi h p://c ea i ecommons.o g/ licenses/by/4.0/. © The Au ho (s) 2021 NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-020-20584-4 ARTICLE NATURE COMMUNICATIONS | (2021) 12:518 | h ps://doi.o g/10.1038/s41467-020-20584-4 | www.na u e.com/na u ecommunica ions 7