Pinpointing cell identity in time and space

molb-07-00209 Augus 13, 2020 Time: 17:7 # 1 PERSPECTIVE published: 14 Augus 2020 doi: 10.3389/ molb.2020.00209 Edi ed by: Eleono a Leucci, KU Leu en, Belgium Re iewed by: Sil ia Bo ini, Uni e si é Cô e d’Azu , F ance S ephen J. Bush, Uni e si y o Ox o d, Uni ed Kingdom Da id Hume, The Uni e si y o Queensland, Aus alia Pa ik S åhl, KTH Royal Ins i u e o Technology, Sweden *Co espondence: Anca F. Sa ulescu [email p o ec ed]; [email p o ec ed] Special y sec ion: This a icle was submi ed o P o ein and RNA Ne wo ks, a sec ion o he jou nal F on ie s in Molecula Biosciences Recei ed: 19 Feb ua y 2020 Accep ed: 30 July 2020 Published: 14 Augus 2020 Ci a ion: Sa ulescu AF, Jacobs C, Negishi Y, Da ignon L and Mhlanga MM (2020) Pinpoin ing Cell Iden i y in Time and Space. F on . Mol. Biosci. 7:209. doi: 10.3389/ molb.2020.00209 Pinpoin ing Cell Iden i y in Time and Space Anca F. Sa ulescu1*, Ca on Jacobs1,2,3, Yu aka Negishi1, Lau ianne Da ignon1and Musa M. Mhlanga1,3,4 1Di ision o Chemical, Sys ems & Syn he ic Biology, Facul y o Heal h Sciences, Ins i u e o In ec ious Disease & Molecula Medicine, Uni e si y o Cape Town, Cape Town, Sou h A ica, 2SAMRC/NHLS/UCT Molecula Mycobac e iology Resea ch Uni , Depa men o Pa hology, Ins i u e o In ec ious Disease and Molecula Medicine, Uni e si y o Cape Town, Cape Town, Sou h A ica, 3Wellcome Cen e o In ec ious Diseases Resea ch in A ica, Uni e si y o Cape Town, Cape Town, Sou h A ica, 4Ins i u o de Medicina Molecula , Faculdade de Medicina da Uni e sidade de Lisboa, Lisbon, Po ugal Mammalian cells display a b oad spec um o pheno ypes, mo phologies, and unc ional niches wi hin biological sys ems. Ou unde s anding o mechanisms a he indi idual cellula le el, and how cells unc ion in conce o o m issues, o gans and sys ems, has been g ea ly acili a ed by cen u ies o ex ensi e wo k o classi y and cha ac e ize cell ypes. Classic his ological app oaches a e now complemen ed wi h ad anced single-cell sequencing and spa ial ansc ip omics o cell iden i y s udies. Eme ging da a sugges s ha addi ional le els o in o ma ion should be conside ed, including he subcellula spa ial dis ibu ion o molecules such as RNA and p o ein, when classi ying cells. In his Pe spec i e piece we desc ibe he impo ance o in eg a ing cell ansc ip ional s a e wi h issue and subcellula spa ial and empo al in o ma ion o ho ough cha ac e iza ion o cell ype and s a e. We e e o ecen s udies making use o single cell RNA-seq and/o image-based cell cha ac e iza ion, which highligh a need o such in-dep h cha ac e iza ion o cell popula ions. We also desc ibe he ad ances equi ed in expe imen al, imaging and analy ical me hods o add ess hese ques ions. This Pe spec i e concludes by aming his a gumen in he con ex o p ojec s such as he Human Cell A las, and ela ed ields o cance esea ch and de elopmen al biology. Keywo ds: spa io empo al localiza ion, cell sub ype classi ica ion, spa ial ansc ip omics, MRNA subcellula localiza ion, cell sub ype INTRODUCTION Biology inhe en ly equi es classi ica ion o manage as amoun s o i educibly complex in o ma ion. Biological sys ems a e b oken up in o o gans, issues, cells and molecula pa hways, whe e cells make up he smalles unc ional uni s o li e. Cells mus hus occupy a wide ange o pheno ypes and mo phologies, ul illing he unc ional equi emen s o each o hei issues and niches. Ou unde s anding o his di e si y is acili a ed by classi ying cells ac oss his spec um as di e en cell “ ypes” ca ying speci ic molecula signa u es. Howe e , cells also exis in dynamic s a es wi h some unc ional plas ici y, which p esen s pa icula challenges o a educ ionis classi ica ion app oach. The e a e wo ypes o classi ica ion e o s we a e a isk o making: ei he assigning he same iden i y o wo cells when hey a e di e en , o con e sely, labeling wo cells as di e en when hey a e unc ionally iden ical. To a oid hese e o s, we need o be e unde s and he pa ame e s ha F on ie s in Molecula Biosciences | www. on ie sin.o g 1Augus 2020 | Volume 7 | A icle 209 molb-07-00209 Augus 13, 2020 Time: 17:7 # 2 Sa ulescu e al. Pinpoin ing Cell Iden i y in Time and Space dis inguish cells, including he ela ionship be ween cell s a e, unc ion, and iden i y, o be able o delinea e cell sub- ypes and classes wi h g ea e esolu ion. To illus a e ou a gumen , conside a hypo he ical si ua ion: wo cells a e adjacen o one ano he in a issue sample and possess simila le els o he same RNA ansc ip s. Does his conclude ha hey a e he same cell ype? In e sely, i hey ha e di e ing le els o ansc ip s, does his mean hey a e di e en , o could hey simply be in di e en s a es o s ages in a p ocess? Cells exis in lux, ac oss con inuous spec a o di e en ia ion and s a e. They p og ess h ough i e e sible p ocesses, such as de elopmen o di e en ia ion; oscilla o y p ocesses such as he cell cycle and ci cadian hy hms; as well as e e sible ansi ions be ween s a es, including nu i ional o disease s a us. These changes na u ally mani es in a cell’s beha io , molecula composi ion and subcellula o ganiza ion o a ious componen s. I is in ui i e o unde s and p ocesses such as de elopmen o di e en ia ion as p og essions o cell ype o iden i y. Howe e , i is less clea i cells in dis inc bu ansi o y s a es should be assigned o dis inc sub-ca ego ies. As a co olla y, by using cu en echniques ha p ima ily conside cells’ molecula composi ion (and e en hei o ganiza ion wi hin a issue) we canno clea ly de e mine whe e in a p ocess a gi en cell may be, and hus how simila o dis inc one cell iden i y may be om ano he . SINGLE-CELL GENOMICS AS THE STANDARD APPROACH TO IDENTIFY CELL TYPES AND STATES A cell’s iden i y is de e mined by i s lineage, p esen s a e, and u u e di e en ia ion o unc ional po en ial, as well as i s spa ial con ex wi hin a issue o sys em (Wagne e al., 2016). Based on he unde s anding ha his iden i y is e lec ed in he molecula composi ion o he cell, single-cell genomics and p o eomics ha e become s anda d app oaches o cha ac e ize cell iden i y and s a e. Single-cell genomics includes measu ing gene exp ession, ypically ia RNA sequencing (RNA-seq), as well as epigene ic s a es and ch oma in s uc u e, using app oaches which ha e been adap ed o wo k a he single-cell le el ( e iewed in T apnell, 2015;Ludwig and Bin u, 2019;Shema e al., 2019). Single-cell RNA-seq (scRNA-seq) has been apidly de eloped and is he mos popula app oach o iden i y cell ypes as i enables classi ica ion o cells unbiasedly, based on gene exp ession pa e ns and allows o iden i y no el cell ypes and sub ypes wi hou p io knowledge ( o example Da manis e al., 2015; G ün e al., 2015;Shekha e al., 2016;Villani e al., 2017; Sch oede e al., 2020). Typical cell- ype iden i ica ion by scRNA- Seq in ol es dissocia ing single cells om issues, ollowed by isola ion o hei RNA, e e se ansc ip ion, ampli ica ion, sequencing and compu a ional analysis (Figu e 1A). Howe e , he e a e se e al limi a ions o scRNA-Seq me hods: (1) ade- o be ween he numbe o cells and da a quali y; (2) limi ed measu emen o p o ein exp ession; (3) noise le el and; (4) lack o spa ial and empo al in o ma ion. As echnical de elopmen s in scRNA-Seq me hodology, including compu a ional me hods a e well documen ed in o he e iews (Hwang e al., 2018;Chen G. e al., 2019;Liao e al., 2020), we discuss ep esen a i e scRNA- Seq me hods and desc ibe hei ad an ages and limi a ions below. TRADE-OFF BETWEEN NUMBER OF THE CELLS AND DATA QUALITY Cu en scRNA-Seq echniques can be classi ied based on he single cell cap u e me hod: low cy ome y [e.g., Sma -Seq2, (Picelli e al., 2013)], mic o luidics [e.g., C1 CAGE (Kouno e al., 2019)], d ople s [e.g., 10x Ch omium, D oNc-seq (Habib e al., 2017)], D op-Seq (Macosko e al., 2015), mic owell [e.g., Seq Well (Gie ahn e al., 2017)], and indexing me hods [e.g., sciRNA- Seq (Cao e al., 2017)]. Flow cy ome y and mic o luidics-based me hods enable us o ob ain addi ional biologically ele an da a o he han gene le el exp ession. Fo example, Sma - seq de ec s iso o m-le el exp ession and mu a ions in exon egions, and C1 CAGE quan i ies e en non-polyadenyla ed RNA such as enhance RNAs. Mul iomics analysis echniques ha e been de eloped o low cy ome y-based me hods. Fo example, scDam&T-seq can analyze RNA le el and p o ein binding si es simul aneously (Rooije s e al., 2019). Howe e , hese me hods can only measu e a maximum amoun o a ew hund ed cells pe expe imen . As such, hey a e no sui able o cha ac e iza ion o a e cell popula ions, as a la ge numbe o cells needs o be analyzed on a single cell le el in hese cases. D ople s, mic owell and indexing me hods sequence only pa s o RNA molecules (in mos cases only he 3’ end o he RNA) e e sed ansc ibed by oligo dT. Thus hese me hods a e unable o measu e iso o m- le el exp ession and non-poly(A)-con aining RNAs. Howe e , hese me hods can measu e o e 1000 cells pe expe imen and, as such, a e sui able o de ec a e cell popula ions. LIMITED MEASUREMENT OF PROTEIN EXPRESSION A ew ecen scRNA-Seq echniques a e able o measu e p o ein exp ession by using oligo conjuga ed an ibodies. Fo example, CITE-Seq uses oligo-conjuga ed an ibodies agains cell su ace ma ke s o quan i y p o ein exp ession le els (S oeckius e al., 2017). Howe e , cu en ly, CITE-seq allows he measu emen o a limi ed numbe o p o eins on he cell su ace. Measu ing exp ession o in acellula p o eins by sequencing-based me hods emains challenging. NOISE LEVEL Compa ed o bulk RNA-Seq, scRNA-seq da a is in insically noisy and highly spa se especially due o so-called d opou e en s - cases in which genes a e no de ec ed despi e being exp essed. Technical a iabili y accoun s o app oxima ely 50% o cell-cell a ia ion in exp ession measu emen s and a ec s downs eam analyses such as clus e ing and pseudo ime econs uc ion. In ac , a la ge ac ion o s ochas ic allele-speci ic exp ession can F on ie s in Molecula Biosciences | www. on ie sin.o g 2Augus 2020 | Volume 7 | A icle 209 molb-07-00209 Augus 13, 2020 Time: 17:7 # 3 Sa ulescu e al. Pinpoin ing Cell Iden i y in Time and Space FIGURE 1 | Po en ial di e en ial spa ial dis ibu ion and clus e ing beha io o RNA ansc ip s in cells, which sha e he same iden i y based on single-cell RNA sequencing. (A) Single-cell RNA-seq wo k low, which ypically yields -SNE plo s shown in (B). B Cells ha a e classi ied as belonging o he same sub ype/g oup based on RNA ansc ip coun migh di e in he subcellula localiza ion o a ious RNA ansc ip s. (C) Spa ially con ined RNA ansc ip s may exhibi non-clus e ed spa ial dis ibu ion o localize in clus e s. (D) Spa ial subcellula localiza ion o RNA ansc ip s may be co ela ed wi h o independen o speci ic cellula s uc u es, o ganelles o ma ke s. be explained by echnical noise, especially o genes exp essed a low and mode a e le els (Kim e al., 2015). Al hough spike- in con ols - syn he ic nucleic acids used o e o calib a ion ha aid in co ec ing noise, can be used (Lun e al., 2017), hese con ols a e no applicable o d ople -based echniques. Se e al me hods ha e been de eloped o denoise da a by compu a ionally impu ing missing alues, including MAGIC, SAVER, scImpu e, DeepImpu e and o he s (Huang M. e al., 2018;Li and Li, 2018; an Dijk e al., 2018;A isdakessian e al., 2019;Luecken and Theis, 2019). S ill, i emains challenging o dis inguish echnical d opou e en s om biological e en s. Addi ionally, i is impo an o eg ess ou biological e en s ha a e no o in e es . Fo example, a no el T cell popula ion was iden i ied only a e emo al o gene exp ession go e ned by he cell cycle (Bue ne e al., 2015). LACK OF SPATIAL AND TEMPORAL INFORMATION S anda d scRNA-Seq me hods canno accoun o spa ial o empo al in o ma ion, wi h some o he main limi a ions being he equisi e cell-dissocia ion s ep, which dis up s he mic oen i onmen and des oys all spa ial ela ionships be ween F on ie s in Molecula Biosciences | www. on ie sin.o g 3Augus 2020 | Volume 7 | A icle 209 molb-07-00209 Augus 13, 2020 Time: 17:7 # 4 Sa ulescu e al. Pinpoin ing Cell Iden i y in Time and Space cells, and he cell-lysis s ep, du ing which spa ial in o ma ion a he subcellula le el is los . Addi ionally, his la e s ep induces a i ac s and migh dis o cell ype iden i ica ion (Adam e al., 2017). Recen a emp s o add ess spa ial and empo al aspec s by RNA-seq ha e eme ged. The i s a emp a p ese ing spa ial in o ma ion in single cell RNA-Seq used in si u ampli ica ion by padlock p obe and RNA sequencing by liga ion (Ke e al., 2013). In a me hod dubbed FISSEQ, Lee e al. (2015) con e ed RNA in ixed cells and issues in o c oss- linked cDNA amplicons, ollowed by manual sequencing on a con ocal mic oscope. This allowed o en ichmen o con ex - speci ic ansc ip s, while p ese ing issue and cell a chi ec u e. While in si u RNA-Seq echniques p o ide he exp ession da a o highly mul iplexed genes wi h high spa ial esolu ion, analysis o he whole ansc ip ome emains challenging. On he o he hand, non-in si u spa ial sequencing echniques ha e been de eloped. “Spa ial ansc ip omics” (ST) (S åhl e al., 2016) and high densi y spa ial ansc ip omics (HDST) (Vicko ic e al., 2019) make use o a slide p in ed wi h an a ay o e e se ansc ip ion oilgo(dT) p ime s, o e which a issue sample is laid. This allows o imaging, ollowed by in si u un a ge ed cDNA syn hesis and RNA-seq. Read coun s can be co ela ed back o he mic oa ay spo and loca ion wi hin he sample. This has a 2D spa ial esolu ion o ∼100 and 2 µm (o se e al cells, and less han 1 cell) pe spo in ST and HDST, espec i ely. The ST echnique is now comme cialized as Visium om 10X genomics. Rod iques e al. (2019) sough o add ess he ques ion o cell-scale spa ial esolu ion in a issue by de eloping SlideSeq. This me hod unc ions by ans e ing RNA om issue sec ions on o a su ace co e ed in DNA- ba coded beads wi h known posi ions. The posi ional sou ce o he RNA wi hin he issue can hen be deduced by sequencing. In addi ion o a ay-based app oaches, a ew pionee ing me hods ha e been de eloped o ob ain spa ial in o ma ion a cell-cell in e ac ions by compu a ional in e ence, physical sepa a ion by lase mic odissec ion and gen le issue dissocia ion (Sa ija e al., 2015;Moo e al., 2018;Giladi e al., 2020). By combining in si u hyb idiza ion images, Sa ija e al. in e ed cellula localiza ion compu a ionally. Al hough his app oach is widely applicable, i is challenging o apply o issues whe e he spa ial pa e n is no ep oducible, such as in a umo , o issues whe e cells wi h highly simila exp ession pa e ns a e spa ially sca e ed ac oss he issue. While mic odissec ion app oaches achie e highe spa ial esolu ion compa ed o a ay-based echniques such as Slide-Seq, hese app oaches only wo k when he sou ce o spa ial a iabili y has a cha ac e is ic mo phological co ela e. Giladi e al. (2020) in oduces a new me hod, PIC-seq, which combines cell so ing o physically in e ac ing cells (PICs) wi h single-cell RNA sequencing and compu a ional modeling o cha ac e ize cell-cell in e ac ions and hei impac on gene exp ession. This app oach has a ew limi a ions: double s migh cause mis-iden i ica ion o cell-cell in e ac ion, and i is no sui able o use on in e ac ing cells ha ha e simila exp ession p o iles. While hese non-in si u echniques can achie e highe de ec ion sensi i i y han in si u RNA-Seq a single-cell o nea ly single-cell esolu ion, we sugges ha u he p ecise spa ial in o ma ion o RNAs and p o eins in he cell is equi ed o ully unde s and cell s a e, as exempli ied by P g anules (see sec ion “Discussion” below). To unde s and he ansi ion be ween cell s a es and di e en ia ion s ages, empo al analyses o he ansc ip ome and epigenome a e essen ial. The majo i y o sequencing- based app oaches p o ide only a “snapsho ” pe spec i e o any sample, and do no allow us o place he in o ma ion in he empo al con ex . To add ess his limi a ion, o e 70 me hods o econs uc pseudo ime ha e been de eloped (Re iewed in Saelens e al., 2019;G ün and G ün, 2020), allowing o he cha ac e iza ion o biological p ocesses’ dynamics mo e accu a ely han con en ional ime se ies o bulk RNA-Seq (T apnell e al., 2014;Ji and Ji, 2016;Reid and We nisch, 2016; Qiu e al., 2017;Chen Y. e al., 2019). Fo example, Monocle (T apnell e al., 2014), uses single-cell RNA-seq da a collec ed a mul iple ime poin s o cha ac e ize he empo al aspec o gene exp ession. This was used o cha ac e ize di e ences in gene exp ession in di e en ia ion o p ima y human myoblas s (T apnell e al., 2014). TSCAN uses RNA-seq da a o compu a ionally o de cells in a he e ogenous popula ion based on he g adual ansi ion o hei gene exp ession (Ji and Ji, 2016). Addi ionally, SPRING is able o isualize long con inuous gene exp ession opologies, ep esen ing a powe ul ool o isualize complex di e en ia ion p ocesses such as b anching opology o hema opoie ic p ogeni o cells and hei di e en ia ion (Wein eb e al., 2018). Ano he ecen ly de eloped me hod, RNA eloci y, econs uc s ajec o y based on kine ics o nascen and ma u e mRNA o mo e solid quan i a i e ounda ion (La Manno e al., 2018). These pseudo ime econs uc ion app oaches can aid in un eiling bo h ansi ions in gene exp ession and he dynamics o ansc ip ional egula ion o cha ac e iza ion o gene egula o y ne wo ks. The elucida ion o gene egula o y ne wo ks can enhance ou unde s anding o complex cellula p ocesses in li ing cells as a b idge connec ing geno ypes and pheno ypes. T adi ional app oaches o ansc ip ome p o iling ha e been success ully used o in e and cha ac e ize egula o y ne wo ks o e ime cou ses such as in di e en ia ion. A no able example is FANTOM5 phase2, which e ealed gene egula o y ne wo ks by dense ime cou se analyses wi h CAGE (A ne e al., 2015; Baillie e al., 2017). Ne wo k cons uc ion om econs uc ed pseudo ime in scRNA-Seq is challenging due o a combina ion o biological a ia ion (e.g., s ochas ici y o bu s s) and echnical limi a ions, such as he inabili y o cap u e non-poly(A) RNAs. To da e, he e a e only small-scale e o s o de i e egula o y ne wo ks om single-cell ansc ip omics da a o e ime cou ses. In p inciple, pseudo ime econs uc ed om scRNAseq da a allows in e ence o gene- egula o y ne wo ks (Aiba e al., 2017). A ecen s udy showed ha combining me hods o ne wo k econs uc ion wi h RNA eloci y imp o es he accu acy o ne wo k in e ence, hus imp o ing he empo al coupling measu emen o mo e accu a e econs uc ion (Qiu e al., 2020). Howe e , as pseudo ime econs uc ion me hods a e based on he assump ion ha changes in gene exp ession a e con inuous o g adual, he app oach is no able o cap u e ei he d as ic o ansien changes in ansc ip ion ha may occu du ing a p ocess o in e es . In addi ion, we ha e o keep in mind ha F on ie s in Molecula Biosciences | www. on ie sin.o g 4Augus 2020 | Volume 7 | A icle 209 molb-07-00209 Augus 13, 2020 Time: 17:7 # 5 Sa ulescu e al. Pinpoin ing Cell Iden i y in Time and Space scRNA-Seq me hods, especially d ople based echniques, migh cap u e double s, which unless iden i ied and emo ed, migh be mis aken as ansien cell s a es (Kisele e al., 2019). As wi h con en ional RNA-seq, hese app oaches a e also unable o ake in o conside a ion he spa ial dis ibu ion o RNAs, despi e he impo ance o changes in RNA subcellula dis ibu ion du ing p ocesses such as de elopmen and di e en ia ion. SPATIAL AND TEMPORAL INFORMATION CAN INFORM A MORE IN-DEPTH SUB-CLASSIFICATION OF CELLS Al hough he e is apid and ongoing de elopmen o sequencing-based echnologies, hei capabili ies o de e mine high- esolu ion spa ial and empo al in o ma ion a e limi ed. Eme ging e idence indica es ha , a a gi en poin in ime, no only RNA and p o ein abundance (Co e e al., 2016), bu also di e en ial subcellula dis ibu ion o hese molecules con ibu es o a cell’s s a e and unc ion ( o example: Moo e al., 2017). Conside once mo e ou hypo he ical si ua ion: can we conclude ha wo cells, adjacen o one ano he in a issue, which possess simila le els o he same RNA ansc ip s, a e he same cell ype (Figu e 1B). By inc easing he spa ial esolu ion a which we assess hese samples, we may obse e ha al hough a simila concen a ions, a pa icula RNA species could be di e en ially localized in hese cells. Fo ins ance, he RNA may be dispe sed ac oss he cy oplasm in one cell and locally clus e ed in he o he (Figu es 1B,C). The subcellula dis ibu ion o mRNA ansc ip s can de e mine hei binding pa ne s and in luence hei a e o ansla ion, a ec ing he cellula concen a ion and localiza ion o he p o ein p oduc (Ka z e al., 2012, 2016;Moo e al., 2018). This, in u n, can in luence he cell’s unc ion and capaci y o espond o a ious en i onmen al cues. Addi ionally, we may obse e spa ial posi ioning o ce ain ansc ip s in close associa ion wi h subcellula landma ks o o ganelles (Sa ulescu e al., 2019, e iewed in Su e , 2018;Hughes and Simmonds, 2019 and o he s) (Figu e 1D), including memb aneless o ganelles such as s ess g anules ( o example Khong e al., 2017;Pad ón e al., 2019;Wilbe z e al., 2019). This could indica e a unc ional ela ionship be ween he RNA’s cellula ole and b oade cellula p ocesses, such as cell di ision, di e en ia ion, pola iza ion e c.’ Fu he , his may also in luence, o be in luenced by, cell s a e o iden i y. The signi icance o RNA/p o ein subcellula dis ibu ion o e bo h space and ime can be illus a ed using he example o P g anules in C. elegans de elopmen (B angwynne e al., 2009). Upon pola iza ion o a C. elegans single-cell emb yo along he an e io -pos e io axis, hese RNA- and p o ein-con aining condensa es shi om a uni o m dis ibu ion o localize a he pos e io hal o he cell. This di e en ial dis ibu ion de e mines he ge m line and soma ic cell a es o he di iding emb yo’s daugh e cells. Thus, al hough he unc ion o P g anules is no ully unde s ood ye , he spa ial dis ibu ion o hese g anules ep esen s a cell-s a e ansi ion ma ke ( he eadiness o he cell o p og ess o he wo-cell s age) as well as a ma ke o de e mina ion o he cell a e (p ogeni o ge m cell and soma ic sis e cell). Impo an ly, applica ion o cu en single- cell sequencing and pseudo ime econs uc ion me hods in his sys em would no ully e eal he ansi ion o cell s a e o ype, as changes in e ed by RNA and p o ein spa ial localiza ion in P g anules and possibly addi ional s uc u es would no be de ec ed. Simila p ocesses in ol ing localiza ion o mRNA ansc ip s occu in o he de elopmen al sys ems, including de e mina ion o spa ial pa e ning in he de eloping D osophila emb yo (Johns one and Lasko, 2001) and de e mina ion o cell a e in he Xenopus oocy e (King e al., 2005). Such sys ems highligh a need o echnologies capable o accoun ing o bo h he empo al and spa ial aspec s o single-cell genomics o unde s anding cell ypes and s a es. In a manne simila o mRNA, o he species o RNA may be subjec o such spa ial o ganiza ion. Ou unde s anding o long non-coding RNA (lncRNA) unc ion and beha io is s ill in i s in ancy. Howe e , he in e sec ion o he subcellula o ganiza ion and unc ion o long non-coding RNAs may con ibu e o a ine classi ica ion o cellula iden i ies and p o e o be a pa icula ly in e es ing ield o disco e y in he u u e. O e all, subcellula RNA dis ibu ion, as well as in e ac ions be ween ansc ip s and cellula s uc u es o a icking and packaging, may di e be ween o he wise simila cells. I hese di e ences lead o unc ional dis inc ions be ween he cells, can we s ill conside hem o ha e he same iden i y? Expanding on his pe spec i e, po en ially hund eds o RNA ansc ip s in a gi en expe imen may be p esen a he same le el be ween cells; howe e hese ansc ip s migh be di e en ially dis ibu ed wi hin hem. Thus a ma ix o housands o po en ial combina ions o RNA localiza ion pa e ns may exis , sugges ing he possibili y o a la ge a ay o g anula ly di e en ia ed cell sub ypes ha migh ha e p e iously been classi ied as belonging o he same g oup. While he discussion he e has ocused on cell classi ica ion by ansc ip ional da a, i is impo an o ecognize ha p o eomic da a and he spa ial o ganiza ion o a cell’s p o ein epe oi e could po en ially con ibu e o cell classi ica ion decisions. Recen la ge scale s udies indica e ha cells ha appea gene ically iden ical display a ious p o ein le els and subcellula localiza ions o p o eins (Sigal e al., 2006;B eke e al., 2013;Thul e al., 2017;Lu e al., 2018) du ing di e en ia ion (Balázsi e al., 2011;Rubakhin e al., 2013), in esponse o en i onmen al s imuli (Na ayanaswamy e al., 2009;Balázsi e al., 2011;B eke e al., 2013) o d ug ea men (Tkach e al., 2012;Déne aud e al., 2013;To es e al., 2016;I zhak e al., 2017;Sha e e al., 2017). This could be due o local ansla ion o di e en ially dis ibu ed mRNAs, o pos - ansla ional modi ica ions o he p o ein, in e ing di e en ial in e ac ions wi h binding pa ne s. Simila ly o RNA, his phenomena may apply o mul iple di e en p o ein species. In eg a ed wi h he spa ial dis ibu ion in o ma ion o RNAs, his could exponen ially expand he ma ix o cellula o ganiza ions, highligh ing he po en ial o in-dep h cellula classi ica ion o accu a ely esol ing cell s a es and iden i ies. The classic app oach o s udy spa ial p o eomics elies on subcellula ac iona ion o o ganelles coupled wi h mass spec ome y analysis. Addi ional spa ial in o ma ion a F on ie s in Molecula Biosciences | www. on ie sin.o g 5Augus 2020 | Volume 7 | A icle 209 molb-07-00209 Augus 13, 2020 Time: 17:7 # 6 Sa ulescu e al. Pinpoin ing Cell Iden i y in Time and Space high esolu ion can be ob ained om app oaches combining p oximi y labeling media ed by enginee ed asco bic acid pe oxidase (APEX) o an ibody-media ed a ini y pu i ica ion wi h mass spec ome y ( o example Hein e al., 2015;Hu lin e al., 2017;Lobingie e al., 2017;Paek e al., 2017). This allows o he cha ac e iza ion o he in e ac ome o a p o ein, based on he assump ion ha p o eins mus be in close p oximi y o be able o in e ac . As such, i is also indica i e o he local spa ial p o eome. Howe e , hese app oaches a e s ill in hei in ancy, and in-dep h co e age o he cell p o eome has no been comple ed o da e (Re iewed in Lundbe g and Bo ne , 2019). A complemen a y app oach o mass spec ome y-based me hods elies on imaging o p o eins on a p o eome-wide le el and a single-cell esolu ion (Re iewed in Lundbe g and Bo ne , 2019). The Human P o ein A las (HPA) ini ia i e aims o map he spa ial subcellula dis ibu ion o all human p o eins in all cell ypes o he human body (Thul e al., 2017; Uhlen e al., 2010;Uhlén e al., 2015;Uhlen e al., 2017). The spa ial dis ibu ion o an ex ensi e numbe o p o eins has been de e mined, using an ibody labeling, con ocal mic oscopy, and manual and compu a ional image analysis, allowing he de ailed classi ica ion o subcellula localiza ion o hese p o eins. Ongoing esea ch is being conduc ed o u he cha ac e ize he spa ial dis ibu ion o p o eins, making use o echnological ad ances in mul iplexed imaging, endogenous p o ein agging, au oma ed luo escence mic oscopy and image analysis ools, including deep neu al ne wo ks (Re iewed in Lundbe g and Bo ne , 2019). I can easonably be expec ed ha simila la ge-scale app oaches could be applied o he s udy o RNA subcellula localiza ion, and allow he cha ac e iza ion o he spa ial dis ibu ion o po en ially all cellula RNAs in cell lines and issues. This would aid in he ine g ained subclassi ica ion o cell ypes and s a es. ANALYTICAL AND IMAGING-BASED METHODS REQUIRED TO ANALYZE SPATIAL INFORMATION Gi en he complexi y o subcellula o ganiza ion, and a cell’s inhe en s a e o lux, we an icipa e ha in-dep h cha ac e iza ion o he subcellula o ganiza ion o mul iple molecula species ac oss la ge numbe s o cells will equi e ad ances in analy ical imaging-based me hods. Such me hods would need (1) he capaci y o label mul iple RNA ansc ip s and p o eins in a mul iplexed manne ; (2) acquisi ion o da a a bo h high spa ial esolu ion and high h oughpu ; and (3) compu a ional amewo ks o quan i a i e image analysis o la ge, mul i-dimensional imaging da a se s. The la e would be pa icula ly impo an o dis inguish be ween he sub le di e ences in spa ial dis ibu ion o molecules which could occu be ween cells. Fu he , hese ools need o be adap able be ween cul u ed cell monolaye s, la ge 3D cul u es, including sphe oids and o ganoids, and in ac issue samples. This would ensu e he cap u e o spa ial in o ma ion unde con olled condi ions and om cells in hei na i e issue con ex . Se e al labeling and imaging modali ies ha e been de eloped o mee condi ions (1) and (2). E o s o inc ease he labeling sensi i i y and h oughpu capaci y o hyb idiza ion-based echniques ha e led o he eme gence o se e al sophis ica ed RNA FISH ( luo escen in si u hyb idiza ion) echniques, in some cases pai ed wi h a ge ed in si u cDNA syn hesis and sequencing. Con en ional single molecule FISH (smFISH) makes use o mul iple sho single-s anded DNA oligonucleo ide p obes, each labeled wi h a single luo opho e, o a ge and speci ically label mRNA (Raj e al., 2008). A key adap a ion o inc eased RNA FISH labeling capaci y has been he use o sequen ial ounds o mul i-colo labeling and imaging o he same sample. An in ui i e a ia ion o his app oach is massi ely mul iplexed cyclic smFISH, such as osmFISH (Codeluppi e al., 2018), which was used o label 33 a ge ed gene ansc ip s o e 13 ounds o labeling o map he cellula a chi ec u e o he mouse neu al co ex. Labeling capaci y is u he boos ed by he adop ion o FISH p obe ba coding app oaches, along wi h sequen ial labeling, as demons a ed wi h mul iplexed e o - obus FISH (MERFISH) (Chen e al., 2015;Mo i and Zhuang, 2016) and sequen ial FISH (seqFISH and seqFISH +) (Lubeck e al., 2014;Eng e al., 2019). These echniques begin o app oach ull- ansc ip ome imaging, wi h he capaci y o label 100 o 10000 s o RNA species a single cell, subcellula (Lubeck e al., 2014;Chen e al., 2015) o e en sub-di ac ion (Eng e al., 2019) esolu ion. STARmap (Wang e al., 2018) uses in si u ampli ica ion o a ge -speci ic p obe ba code egions, ha can be decoded by 3D sequencing wi hin samples con e ed o a hyd ogel ma ix. This allows o he de ec ion o 1000 s o RNA species in la ge cell numbe s in 3D issue s uc u es. While hese app oaches can all p o ide g ea insigh in o he unc ional cellula o ganiza ion wi hin issues, hey each ha e a ying limi a ions in he spa io empo al esolu ion o h oughpu a ailable. Many o hese app oaches bene i om specialized LabWa e and equipmen . The e is an inc easing ease o access o a o dable liquid handling sys ems, d i en by open we -lab solu ions such as OpenLH (Gome e al., 2019) and modula Lego-based and 3D-p in ed injec ion pumps ( o example Almada e al., 2019). Such sys ems a e essen ial o high-cycle sequen ial labeling o ens o hund eds o molecula a ge s, which can be labeled in a single sample wi hou ba coding, as demons a ed wi h osmFISH (Codeluppi e al., 2018). Au oma ed mic oscopes allow inc eased imaging h oughpu o bo h sample size and numbe , howe e his is ypically done a low magni ica ion and esolu ion. High- esolu ion isualiza ion o molecula a ge s is usually limi ed o he single-cell scale. A new imaging modali y e med syn he ic ape u e op ics (i.e., S ella ision Mic oscope, Op ical Biosys ems) (desc ibed in Ryu e al., 2006) uses in e e ome y o inc ease he e ec i e esolu ion o low magni ica ion imaging. This d i es signi ican ly highe h oughpu (∼100 s–1000 s o cells) o high esolu ion (subcellula and up o single- molecule) imaging. Such a sys em, pa icula ly i coupled wi h on-line luid handling o sequen ial labeling, would be well sui ed o quan i a i e spa ial cha ac e iza ion o he molecula epe oi e o la ge numbe s o indi idual cells o issue sec ions. F on ie s in Molecula Biosciences | www. on ie sin.o g 6Augus 2020 | Volume 7 | A icle 209 molb-07-00209 Augus 13, 2020 Time: 17:7 # 7 Sa ulescu e al. Pinpoin ing Cell Iden i y in Time and Space Requi emen (3), he a ailabili y o compu a ional amewo ks o quan i a i e image analysis o la ge imaging da a se s is an a ea o apid ongoing de elopmen . This spans he ull pos acquisi ion pipeline, om basic image p ocessing o spo de ec ion, decoding, quan i ica ion, and classi ica ion. Quan i a i e image analysis ools commonly used o image- based pheno ypic cell p o iling include ways o acili a e ea u e ex ac ion, da a quali y con ol and no maliza ion, dimensionali y educ ion and clus e ing om la ge numbe s o cells (desc ibed in Caicedo e al., 2017). Many o hese echniques could be adap ed o he analysis o spa ial and empo al cha ac e iza ion by ansc ip omics. Wi h he ma u a ion o deep lea ning echnology, he e is also ex ensi e po en ial o he applica ion o a i icial neu al ne wo k-based app oaches o image p ocessing and analysis. Deep lea ning-based app oaches a e ypically bes -sui ed o he p ocessing o la ge se s o complex da a wi h many pa ame e s, as would be expec ed o hese imaging assays. Applica ions o neu al ne wo k-based app oaches include image denoising, segmen a ion and ca ego iza ion. Se e al neu al ne wo k-based ools a e al eady a ailable o he es o a ion o images wi h high noise le els ( o example Weige e al., 2018; Ba son and Roye , 2019;K ull e al., 2019). Tools o ne wo k- based FISH spo de ec ion ha e likewise s a ed o eme ge ( o example Gudla e al., 2017;Mabaso e al., 2018). A common challenge wi h smFISH applica ions is he densi y o signal, pa icula ly in he case o abundan ansc ip s. This is linked o high le els o backg ound signal, comp omising he signal- o-noise a io (SNR) and ou abili y o au oma ically de ec and quan i y spo s. Al eady-a ailable ne wo k-based denoising and spo de ec ion ools could be u he adap ed o he pa icula ly challenging low SNR and high haze condi ions commonly encoun e ed in smFISH. In addi ion o spo de ec ion, spa ially esol ed ansc ip omics necessi a es he abili y o dis inguish indi idual and adjacen cells om each o he , and a way o cha ac e ize he dis ibu ion o FISH spo s ac oss indi idual cells. While he e a e a la ge numbe o cell segmen a ion ools a ailable ( e iewed in Meije ing, 2012 and Vica e al., 2019), he au oma ed segmen a ion o densely packed cells and nuclei, ei he in a cul u ed monolaye o in ac issue sec ions emains a challenge. Po en ial solu ions o his may also lie in machine lea ning and ne wo k-based app oaches ( o example Al-Ko ahi e al., 2018;Schmid e al., 2018;Be g e al., 2019). Beyond image p ocessing, cha ac e iza ion o cellula FISH spo dis ibu ion pa e ns, including quan i ica ion pe cellula compa men , could make use o simila app oaches o hose used o localiza ion pa e n classi ica ion in spa ial p o eomics. These include K-nea es neighbo classi ie s, suppo ec o machines, a i icial neu al ne wo ks and decision ees ( e iewed in Lundbe g and Bo ne , 2019). As expe imen al echnologies de elop and gene a e high- esolu ion spa ial and empo al cell cha ac e iza ion da ase s, ongoing de elopmen o ools and pla o ms o analyze his da a will be impe a i e. Many o he no el image p ocessing and analysis ools desc ibed abo e equi e op imiza ion o high h oughpu . In addi ion, he e is a need o de elopmen o comple e analy ical pipelines and amewo ks o p ocessing and ex ac ing he complex in o ma ion and pa e ns om hese imaging da a se s. Ea ly i e a ions o such amewo ks can be seen in eme ging pla o ms such as DypFISH (Dynamic pa e ned FISH) (Sa ulescu e al., 2019), and S a ish, unde de elopmen by he SpaceTx conso ium in associa ion wi h he Human Cell A las p ojec (desc ibed in Pe kel, 2019). DypFISH is a ecen ly de eloped analy ical pla o m o quan i a i e cha ac e iza ion o he spa ial and empo al subcellula dis ibu ion o key biomolecules a a single cell le el. This sys em makes use o mic opa e ning o cons ain he a chi ec u e o he cell, in e ing a educ ion in a ia ion o subcellula dis ibu ion o mRNA and p o ein and allowing o high ep oducibili y. This app oach was used o quan i y he co ela ion o mRNA and p o ein spa ial dis ibu ions and he MTOC (a key indica o o a cell’s pola i y) in mouse ib oblas s, e ealing impo an spa ial and empo al di e ences be ween mRNA species, as well as wi hin an mRNA species du ing pola iza ion. This may indica e di e en ial cell s a e-dependen spa ial dis ibu ion o impo an biomolecules. DypFISH may hus be a i s s ep in es ablishing a mo e comp ehensi e app oach o he cha ac e iza ion o spa ial and empo al in o ma ion in issues and o he biological sys ems wi h high le els o complexi y. The S a ish pla o m seeks o add ess c i ical aspec s o da a handling and p e-p ocessing, as well as spo de ec ion and RNA iden i ica ion in a lexible manne . This enables he pla o m o handle da a se s om mul iple echniques al eady desc ibed he e, and ex ac and compa e in o ma ion ac oss di e en expe imen s. Mo e ecen ly, Spa ialDB (Fan e al., 2020) has been se up as a manually cu a ed and explo able eposi o y o spa ially esol ed ansc ip omic da ase s om mul iple echniques. As hese analy ical pla o ms de elop, we expec ha he in eg a ion o each o hese ools in o a single amewo k, in a modula manne , will be bene icial o esea che s seeking o unde s and cell iden i y and di e ences in biology. Indeed, he e is al eady a pa allel d i e o he in eg a ion o single cell sequencing app oaches wi h imaging-based app oaches. RNAscope, which makes use o b anched DNA, (Wang e al., 2012, ma ke ed by ACDBio) has been shown o be amenable o mul iplexing and image-based ansc ip omics, especially when pai ed wi h app oaches such as au oma ed liquid handling (Ba ich e al., 2013) and FISH p obe ba coding (Xia e al., 2019). Mo e ecen ly, RNAscope has been pai ed wi h scRNA seq o demons a e molecula he e ogenei y and cellula dynamics in epide mal wound healing (Haensel e al., 2020). scRNA seq, combined wi h bo h ST and a ge ed in si u sequencing, has been used o compile an a las o he de eloping human hea (Asp e al., 2019). In ano he case, ST has been combined wi h scRNA seq o po ions o he same issue sample, o cha ac e ize he issue a chi ec u e in panc ea ic duc al adenoca cinoma (Moncada e al., 2020). Compu a ional app oaches o allow he in eg a ion o hese dis inc ypes o da a a e also apidly de eloping. Moncada e al. (2020) made use o mul imodal in e sec ion analysis o in eg a e he image and sequencing da a. O he ecen ly ad anced analy ical me hods include he use o p obabilis ic models (Ande sson e al., 2019), supe ised lea ning app oaches o mixed-da a decomposi ion (Cable e al., 2020), and SPOTligh (Elosua e al., 2020), which uses non-nega i e F on ie s in Molecula Biosciences | www. on ie sin.o g 7Augus 2020 | Volume 7 | A icle 209 molb-07-00209 Augus 13, 2020 Time: 17:7 # 8 Sa ulescu e al. Pinpoin ing Cell Iden i y in Time and Space ma ix ac o iza ion eg ession models o he decon olu ion o ST spo da a. DISCUSSION The in ica e ela ionship be ween a cell’s subcellula molecula o ganiza ion, i s spa io empo al con ex wi hin a issue and sys em, and i s iden i y and unc ion has a signi ican impac on ou unde s anding o cell biology. The bene i o high- esolu ion spa io empo al cell ype cha ac e iza ion, aking in o accoun issue- o subcellula - scale in o ma ion, is e iden o a numbe o esea ch ields, including de elopmen al biology (Asp e al., 2019), cance esea ch ( o example Baccin e al., 2020; Moncada e al., 2020 and Yoousu e al., 2020), and p ecision medicine (Pe i p ez e al., 2018). Spa io empo al cha ac e iza ion o he umo mic oen i onmen , o example, can p o ide insigh in o he composi ion, o ganiza ion and unc ionali y o umo - associa ed cells, and hei oles in umo de elopmen and se e i y o disease ( e iewed in Pe i p ez e al., 2018). This kind o esea ch also holds he po en ial o mo e sophis ica ed app oaches o ea ing malignan umo s, o example, whe e cells may ha e p e iously unde app ecia ed ansi ional s a es ha can be a ge ed. I may also be in e es ing o assess i di e en ial spa ial and empo al dis ibu ion o key disease bioma ke s, in addi ion o hei exp ession le els, could be linked o a ia ion in esponse o ea men be ween pa ien s. Recen s udies a single-cell esolu ion indica e ha subcellula spa io empo al ansc ip omic cha ac e iza ion could also help us unde s and he molecula basis and p og ession o ce ain gene ic diso de s, such as A hy hmogenic ca diomyopa hy (Booge d e al., 2019) and cogni i e diseases such as Alzheime ’s Disease (Chen W.-T. e al., 2019) and Pa kinson’s Disease (Aguila e al., 2019). The ole o cell s a e is also inc easingly app ecia ed in in ec ion and immuni y ( e iewed in Kunz e al., 2018), pa icula ly in in ec ious diseases whe e immune egula ion is key o disease ou comes. Fo example, he lineage and me abolic s a e o mac ophages can ha e p o ound e ec s in Mycobac e ial ube culosis in ec ion (Huang L. e al., 2018, e iewed in Shi e al., 2019). Full cha ac e iza ion o mac ophage cell ypes and s a es may imp o e ou unde s anding o , and abili y o be e ea , TB disease. T ansc ip ion-based cell classi ica ion is inhe en o ini ia i es such as he Human Cell A las p ojec . He e, oo, he inco po a ion o high- esolu ion spa io empo al in o ma ion holds impo an po en ial o u he biological insigh s and may g ea ly enhance he ansla ional bene i s o hese ini ia i es. Cu en cell classi ica ion p ocesses will need o be adap ed o include his highe g anula i y o in o ma ion, and hese la ge- scale p ojec s can be expec ed o d i e he in eg a ion o no el expe imen al and imaging echnologies o spa io empo ally esol ed cha ac e iza ion. This will include new and ad anced analy ical app oaches and da a ep esen a ion me hods. Such de elopmen s can highligh and make accessible he weal h o in o ma ion a ailable by hese app oaches. Th oughou his Pe spec i e we ha e emphasized he need o ake in o accoun spa io empo al in o ma ion when cha ac e izing cell s a e and iden i y. As we ha e discussed, an inc easing body o da a suppo s he e ec o a ia ion in mRNA/p o ein exp ession and subcellula localiza ion in di ec ing cell s a e and iden i y. Ne e heless, i is impo an o acknowledge ha no all such a ia ion is necessa ily associa ed wi h unc ional changes in cell s a e. T ansc ip ional egula ion, a a single-cell le el in mammalian cells, is p obabilis ic and in e mi en . This leads o p oduc ion o mRNA ansc ip s in pulses and can con ibu e o cell- o-cell he e ogenei y (Femino e al., 1998;Cos elloe e al., 1999, e iewed in Hume, 2000). This, again, poin s o he ques ion o how we delinea e cell s a e and iden i y, using in eg a ed single-cell ansc ip omics and issue-le el and subcellula spa ial o ganiza ion da a. Mul i- scale spa io empo al in o ma ion, o e a la ge numbe o cells and samples, is needed o us o quan i a i ely assess he ex en o a ia ion ac oss issues and wi hin cells, and o de ec a e e en s and cell ypes. Assessmen o such da a should ake place in he con ex o ou inc easing unde s anding o basic in acellula p ocesses, and he unc ions o issues and disease s a es being s udied. These s udies and analyses will be key in ueling impo an discussions wi hin he ield. Pe inen ly, he di e en scales o esolu ion ha may apply o di e en ques ions should be ca e ully conside ed. Single-cell ansc ip omics and subcellula spa io empo al o ganiza ion con ibu e o cell s a e and iden i y, and can con ibu e o issue unc ion. Howe e , his subcellula esolu ion may no always be necessa y o unde s and he ole o indi idual cells and how hey in e ac wi h neighbo ing cells in he con ex o hei issue. A ela ed discussion is necessa y a ound wha “ h eshold” (o mul iple si ua ion-dependen h esholds) o he ex en o spa io empo al a ia ion (de ec ion o which is esolu ion-dependen ) may be conside ed o delinea e unc ionally dis inc s a es o iden i ies o indi idual cells. Accoun ing o he dynamic s a es and unc ional plas ici y a ailable o cells has al eady eme ged as impo an o he classi ica ion and cha ac e iza ion o cell ypes. This will become mo e widely acknowledged wi h he pa allel de elopmen o powe ul ools and echnology o p oduce, p ocess and mine he eme ging in o ma ion. These and o he necessa y de elopmen s desc ibed he e will allow accu a e, high- esolu ion cell classi ica ion and imp o ed unde s anding o he unc ion o di e en cells in issues. Taken oge he , hese ad ancemen s will p o ide powe ul ools o ad ances in undamen al biology, biomedical esea ch and ela ed ields. AUTHOR CONTRIBUTIONS AS, CJ, LD, and MM concei ed he idea. AS, CJ, LD, and YN w o e and p oo ead he manusc ip . All au ho s ead and app o ed he manusc ip . ACKNOWLEDGMENTS We hank Timo hy de We , Dana M. Sa ulescu, Nicolas Beaume, and Ashley J. Jacobs o ui ul discussions and c i ical eading o he manusc ip . F on ie s in Molecula Biosciences | www. on ie sin.o g 8Augus 2020 | Volume 7 | A icle 209 molb-07-00209 Augus 13, 2020 Time: 17:7 # 9 Sa ulescu e al. Pinpoin ing Cell Iden i y in Time and Space REFERENCES Adam, M., Po e , A. S., and Po e , S. S. (2017). Psych ophilic p o eases d ama ically educe single cell RNA-seq a i ac s: a molecula a las o kidney de elopmen . De elopmen 144, 3625–3632. doi: 10.1242/de .151142 Aguila, J., Cheng, S., Kee, N., Cao, M., Deng, Q., and Hedlund, E. (2019). Spa ial ansc ip omics iden i ies no el ma ke s o ulne able and esis an midb ain dopamine neu ons. bioRxi doi: 10.1101/334417 Aiba , S., Gonzáles-Blas, C. B., Moe man, T., Huynh-Thu, V. A., Im icho a, H., Hulselmans, G., e al. (2017). SCENIC: single-cell egula o y ne wo k in e ence and clus e ing. Na . Me hods 14, 1083–1086. doi: 10.1038/nme h.4463 Al-Ko ahi, Y., Zal sman, A., G a es, R., Ma shall, W., and Rusu, M. (2018). A deep lea ning-based algo i hm o 2-D cell segmen a ion in mic oscopy images. BMC Bioin o m. 19:365. doi: 10.1186/s12859-018-2375-z Almada, P., Pe ei a, P. M., Culley, S., Caillol, G., Bo oni-Rueda, F., Dix, C. L., e al. (2019). Au oma ing mul imodal mic oscopy wi h NanoJ-Fluidics. Na . Commun. 10:1223. Ande sson, A., Be gens åhle, J., Asp, M., Be gens åhle, L., Ju ek, A., Na a o, J. F., e al. (2019). Spa ial mapping o cell ypes by in eg a ion o ansc ip omics da a. bioRxi doi: 10.1101/2019.12.13.874495 A isdakessian, C., Poi ion, O., Yuni s, B., Zhu, X., and Ga mi e, L. X. (2019). DeepImpu e: an accu a e, as , and scalable deep neu al ne wo k me hod o impu e single-cell RNA-seq da a. Genome Biol. 20:211. A ne , E., Daub, C. O., Vi ing-See up, K., Ande sson, R., Lilje, B., D abløs, F., e al. (2015). T ansc ibed enhance s lead wa es o coo dina ed ansc ip ion in ansi ioning mammalian cells. Science 347, 1010–1014. doi: 10.1126/science. 1259418 Asp, M., Giacomello, S., La sson, L., Wu, C., Fü h, D., Qian, X., e al. (2019). A spa io empo al o gan-wide gene exp ession and cell a las o he de eloping human hea . Cell 179, 1647–1660. Baccin, C., Al-Sabah, J., Vel en, L., Helbling, P. M., G ünschläge , F., He nández- Malmie ca, P., e al. (2020). Combined single-cell and spa ial ansc ip omics e eal he molecula , cellula and spa ial bone ma ow niche o ganiza ion. Na . Cell Biol. 22, 38–48. doi: 10.1038/s41556-019-0439-6 Baillie, J. K., A ne , E., Daub, C., De Hoon, M., I oh, M., Kawaji, H., e al. (2017). Analysis o he human monocy e-de i ed mac ophage ansc ip ome and esponse o lipopolysaccha ide p o ides new insigh s in o gene ic ae iology o in lamma o y bowel disease. PLoS Gene . 13:e1006641. doi: 10.1371/jou nal. pgen.1006641 Balázsi, G., an Oudenaa den, A., and Collins, J. J. (2011). Cellula decision making and biological noise: om mic obes o mammals. Cell 144, 910–925. doi: 10.1016/j.cell.2011.01.030 Ba son, J., and Roye , L. (2019). “Noise2Sel : blind denoising by sel -supe ision,” in P oceedings o he 36 he In e na ional Con e ence on Machine Lea ning (Long Beach, CA: PMLR). Ba ich, N., S oege , T., and Pelkmans, N. (2013). Image-based ansc ip omics in housands o single human cells a single-molecule esolu ion. Na . Me hods 10, 1127–1133. doi: 10.1038/nme h.2657 Be g, S., Ku a, D., K oege , T., S aehle, C. N., Kausle , B. X., Haubold, C., e al. (2019). ilas ik: in e ac i e machine lea ning o (bio)image analysis. Na . Me hods 16, 1226–1232. doi: 10.1038/s41592-019-0582-9 Booge d, C. J., Lac az, G. P., Vé esy, Á, Pe ini, I., de Rui e , H., B odehl, A., e al. (2019). Spa ial ansc ip omics Un eil ZBTB11 as a egula o o ca diomyocy e degene a ion in a hy hmogenic ca diomyopa hy. Ci c. Res. 125, A510. B angwynne, C. P., Eckmann, C. R., Cou son, D. S., Ryba ska, A., Hoege, C., Gha akhani, J., e al. (2009). Ge mline P g anules a e liquid d ople s ha localize by con olled dissolu ion/condensa ion. Science 324, 1729–1732. doi: 10.1126/ science.1172046 B eke , M., Gym ek, M., and Schuldine , M. (2013). A no el single-cell sc eening pla o m e eals p o eome plas ici y du ing yeas s ess esponses. J. Cell Biol. 200, 839–850. doi: 10.1083/jcb.201301120 Bue ne , F., Na a ajan, K. N., Casale, F. P., P ose pio, V., Scialdone, A., Theis, F. J., e al. (2015). Compu a ional analysis o cell- o-cell he e ogenei y in single-cell RNA-sequencing da a e eals hidden subpopula ions o cells. Na . Bio echnol. 33, 155–160. doi: 10.1038/nb .3102 Cable, D. M., Mu ay, E., Zou, L. S., Goe a, A., Macosko, E. Z., Chen, F., e al. (2020). Robus decomposi ion o cell ype mix u es in spa ial ansc ip omics. bioRxi doi: 10.1101/2020.05.07.082750 Caicedo, J. C., Coope , S., Heigwe , F., Wa chal, S., Qiu, P., Molna , C., e al. (2017). Da a-analysis s a egies o image-based cell p o iling. Na . Me hods 14, 849–863. Cao, J., Packe , J. S., Ramani, V., Cusano ich, D. A., Huynh, C., Daza, R., e al. (2017). Comp ehensi e single-cell ansc ip ional p o iling o a mul icellula o ganism. Science 357, 661–667. doi: 10.1126/science.aam8940 Chen, G., Ning, B., and Shi, T. (2019). Single-cell RNA-Seq echnologies and ela ed compu a ional da a analysis. F on . Gene . 10:317. doi: 10.3389/ gene. 2019.00317 Chen, W.-T., Lu, A., C aessae s, K., Pa ie, B., F ige io, C. S., Mancuso, R., e al. (2019). Spa ial and empo al ansc ip omics e eal mic oglia-as oglia c oss alk in he amyloid-βplaque cell niche o Alzheime ’s disease. bioRxi doi: 10.1101/719930 Chen, Y., Zhang, Y., and Ouyang, Z. (2019). LISA: accu a e econs uc ion o cell ajec o y and pseudo- ime o massi e single cell RNA-seq da a. Pac. Symp. Biocompu . 24, 338–349. Chen, K., Boe inge , J., Mo i , J., Wang, S., and Zhuang, X. (2015). Spa ially esol ed, highly mul iplexed RNA p o iling in single cells. Science 348:aaa6090. doi: 10.1126/science.aaa6090 Codeluppi, S., Bo m, L. E., Zeisel, A., La Manno, G., an Lun e en, J. A., S ensson, C. I., e al. (2018). Spa ial o ganiza ion o he soma osenso y co ex e ealed by osmFISH. Na . Me hods 15, 932–935. doi: 10.1038/s41592-018-0175-z Cos elloe, E. O., S acey, K. J., An alis, T. M., and Hume, D. A. (1999). Regula ion o he plasminogen ac i a o inhibi o -2 (PAI-2) gene in mu ine mac ophages. demons a ion o a no el pa e n o esponsi eness o bac e ial endo oxin. J. Leukoc. Biol. 66, 172–182. doi: 10.1002/jlb.66.1.172 Co e, A. J., McLeod, C. M., Fa ell, M. J., McClanahan, P. D., Dunagin, M. C., Raj, A., e al. (2016). Single-cell di e ences in ma ix gene exp ession do no p edic ma ix deposi ion. Na . Commun. 7:10865. Da manis, S., Sloan, S. A., Zhang, Y., Enge, M., Caneda, C., Shue , L. M., e al. (2015). Single cell analysis o he human b ain. PNAS 112, 7285–7290. Déne aud, N., Becke , J., Delgado-Gonzalo, R., Damay, P., Rajkuma , A. S., Unse , M., e al. (2013). A chemos a a ay enables he spa io- empo al analysis o he yeas p o eome. PNAS 110, 15842–15847. doi: 10.1073/pnas.1308265110 Elosua, M., Nie o, P., Me eu, E., Gu , I., and Heyn, H. (2020). SPOTligh : seeded NMF eg ession o decon olu e spa ial ansc ip omics spo s wi h single-cell ansc ip omes. bioRxi doi: 10.1101/2020.06.03.131334 Eng, C. L., Lawson, M., Zhu, Q., D ies, R., Koulena, N., Takei, Y., e al. (2019). T ansc ip ome-scale supe - esol ed imaging in issues by RNA seqFISH. Na u e 568, 235–239. doi: 10.1038/s41586-019-1049-y Fan, Z., Chen, R., and Chen, X. (2020). Spa ialDB: a da abase o spa ially esol ed ansc ip omes. Nucl. Acids Res. 48, D233–D237. Femino, A. M., Fay, F. S., Foga y, K., and Singe , R. H. (1998). Visualiza ion o single RNA ansc ip s in si u. Science 280, 585–590. doi: 10.1126/science.280. 5363.585 Gie ahn, T. M., Wadswo h, M. H. II, Hughes, T. K., B yson, B. D., Bu le , A., Sa ija, R., e al. (2017). Seq-Well: po able, low-cos RNA sequencing o single cells a high h oughpu . Na . Me hods 14, 395–398. doi: 10.1038/nme h.4179 Giladi, A., Cohen, M., Medaglia, C., Ba an, Y., Li, B., Zada, M., e al. (2020). Dissec ing cellula c oss alk by sequencing physically in e ac ing cells. Na . Bio echnol. 38, 629–637. doi: 10.1038/s41587-020-0442-2 Gome, G., Waksbe g, J., G ishko, A., Wald, I. Y., and Zucke man, O. (2019). “OpenLH: open liquid-handling sys em o c ea i e expe imen a ion wi h biology,” in P oceedings o he Thi een h In e na ional Con e ence on Tangible, Embedded, and Embodied In e ac ion; TEI ‘19 (New Yo k, NY: ACM). G ün, D., Lyubimo a, A., Kes e , L., Wieb ands, K., Basak, O., Sasaki, N., e al. (2015). Single-cell messenge RNA sequencing e eals a e in es inal cell ypes. Na u e 525, 251–255. doi: 10.1038/na u e14966 G ün, S., and G ün, D. (2020). Deciphe ing cell a e decision by in eg a ed single- cell sequencing analysis. Annu. Re . Biomed. Da a Sci. 3, 1–22. doi: 10.1146/ annu e -bioda asci-111419-091750 Gudla, P. R., Nakayama, K., Pego a o, G., and Mis eli, T. (2017). Spo Lea n: con olu ional neu al ne wo k o de ec ion o luo escence in si u hyb idiza ion (FISH) signals in high- h oughpu imaging app oaches. Cold Sp ing Ha b. Symp. Quan . Biol. 82, 57–70. doi: 10.1101/sqb.2017.82.033761 Habib, N., A aham-Da idi, I., Basu, A., Bu ks, T., Shekha , K., Ho ee, M., e al. (2017). Massi ely pa allel single-nucleus RNA-seq wi h D oNc-seq. Na . Me hods 14, 955–958. doi: 10.1038/nme h.4407 F on ie s in Molecula Biosciences | www. on ie sin.o g 9Augus 2020 | Volume 7 | A icle 209