CHOOSE THE MOST TRUSTED LONG-READ TECHNOLOGY FOR YOUR CORE

Sequence with Confidence The Sequel® II and IIe Systems are powered by Single Molecule, Real-Time (SMRT®) Sequencing, a technology proven to produce highly accurate long reads, known as HiFi reads, for sequencing data you and your customers can trust. SMRT SEQUENCING IS SMART BUSINESS HiFi Reads: PacBio is the only sequencing technology to offer highly accurate long reads. Because HiFi reads are extremely accurate, downstream analysis is simplified and streamlined, requiring less compute time than the error-prone long reads of other technologies. High Throughput: The Sequel II and IIe Systems have high data yields on robust, highly automated platforms to increase productivity and reduce project costs.

Efficient and Easy-To-Use Workflows: Our end-to-end solutions feature library preparation in <3 hours and many push- button analysis workflows, so you can run projects quickly and easily.

Support: All of our products are backed by a global team of scientists, bioinformaticians, and engineers who stand ready to provide you with outstanding service.

OUTSTANDING PERFORMANCE AND RELIABILITY

99% of runs on the Sequel II System completed successfully

Sequel II Systems provide reliable performance with the total bases produced by the PacBio fleet steadily increasing, and 99% of runs completed successfully.

“In our experience, the Sequel II System was essentially production-ready right out of the box. We have used it for a range of applications and sample types — from human genome sequencing to metagenome and microbiome profiling to non-model plant and animal genomes — and results have been very good.” — Luke Tallon, Director of the Genomics Resource Center at Maryland Genomics

pacb.com/Sequel SMRT SEQUENCING APPLICATIONS – EFFICIENT AND COST EFFECTIVE The Sequel II and IIe Systems support a wide range of applications, each adding unique value to a sequencing study. The price per sample depends on multiplex level and the service provider’s fee-for-service rate. A typical service provider fee-for-service rate is a 2- to 4-fold increase.

Whole Genome RNA Targeted Microbial Sequencing Sequencing Sequencing Sequencing Applications De novo Variant Transcriptome Amplicon: De novo Metagenome Metagenomic Full-length 16S assembly detection No-Amp sequencing (1-20 kb) assembly assembly profiling rRNA sequencing (2 Gb) (3 Gb)

Generation of Exhaustive Enrichment complete or Complete, Assembly-free Haplotype Recovery of 8-10 High precision isoform of hard-to- near- complete Unambiguous Why Choose contiguous, comprehensive phasing while Closed genomes full-length genes and recall for all characterization amplify regions, genome species or strain- and correct genome detecting all and plasmids per HiFi read, SMRT Sequencing? variant types with no assembly like repeat assemblies level resolution assemblies annotation variant types without assembly required expansions from microbial population

Up to 184,320 Samples/Year Up to 120 Up to 1,920 Up to 11,520 Up to 19,200 Up to 200 Up to 600 Up to 115,200 Up to 240 Up to 240 whole samples for 1 kb samples for tissues samples microbes samples samples samples (Mulitplex/ genomes at 2 Gb transcriptomes amplicons * 3 Gb genomes (8) (48) (48) (1) (3) (192) SMRT Cell 8M) (768)

Reagents Cost/ $1,300 $2,600 $185 $1,300 $1–4 $82–118 $70 $1,300 $450 $7.50 Sample†

*Study design, sample type, and level of multiplexing may affect the number of SMRT® Cells 8M required. †All prices are listed in USD and cost may vary by region. Pricing includes library and sequencing reagents run on your Sequel II or IIe System and does not include instrument amortization or other reagents.

“My sequencing center has seen the level of interest in projects to be run on our Sequel II System increase by over 100% this year, as compared to the previous year. Many of these projects are coming from investigators/collaborators that are new to my center, they are interested in using HiFi sequencing in their research.” — Bruce Kingham, Director, Sequencing and Genotyping, University of Delaware

IMPROVED ECONOMICS WITH THE SEQUEL II SYSTEM When run at full capacity, the Sequel II and IIe Systems can sequence 240 SMRT Cells 8M per year. And the total instrument cost can be recovered in less than three years by running only 12 SMRT Cells 8M per month.

Instrument Payback Period

Usage Levels Time to Payback (in months)

SMRT Cells/ “As a PacBio service provider we see Capacity 3x Markup 3.5x Markup 4x Markup Month high demand for our Sequel II services. On average we run at 100% 60% 12 26 21 18 capacity with an average project queue of 1.5 months or greater” 80% 16 20 16 13 — Edward Wilcox, Manager, DNA Sequencing Center, 100% 20 16 13 11 Brigham Young University

Months of operation required at various capacity levels and consumables markup rates to recoup Sequel II or IIe System cost, 1 FTE for 3 years at $75k/yr, and service contracts.

pacb.com/Applications When Accuracy Matters — Choose HiFi Sequencing

WHAT IS THE TRUE COST PER BASE IF YOU DON’T GET THE ANSWER? HiFi reads achieve a higher quality assembly of the human genome compared to other technology platforms. With fewer sequencing gaps, HiFi reads capture 5% more variants in the medical exome, including 193 medically-relevant genes that are not fully captured with other sequencing methods. When gaps are present and genes are missed, the cost and time to obtain biological conclusions increase greatly and limit scientific progress.

“PacBio Sequel II has been generating amazing data both in size and quality. All the customers have been very pleased with the data they are getting off of PacBio Sequel II” — Nasun Hah, Director, Genomics Sequencing Core, Salk Institute

ARTICLES NATURE BIOTECHNOLOGY ARTICLES ARTICLES NATURE BIOTECHNOLOGY NATURE BIOTECHNOLOGY a Discordances per 100 kb b a 100 b GRCh37 chr15:43,891,619–43,911,196 (19 kb) 1,000 100 10 5 4 3 2 1 0 a b 100 13.5 kb CCS 10043,895,000 43,900,000 43,905,000 43,910,000 GRCh37 100chr15:43,891,619–43,911,196 (19 kb) 2 × 250 bp NGS 99.9% 99 13.5 kb CCS 43,895,000 9043,900,000 43,905,000 43,910,000 2 × 250 bp NGS STRCSTRC gene gene 99 80 80 99.5% 97.0% 98 70 99.0% 2 × 250 bp NGS STRC gene 60 ONT + 60 GRCh38 97 2x250 bp short 98 Illumina PB CLR PB CCSHiFi 2 × 250 bp NGS 50 NGS reads HG002 CCS 40 40 96 97 13.5 kb, 28× 13.5 kb CCS 30 ONT

Predicted NG50 (Mb) 20 CHM1 CLR, 70× 95 96 20 13.5 kb HiFi reads 10

13.5 kb CCS 100 kb chunks (cumulative %) 94 0 Non-gap positions in GRCh37 covered (%) 0204060 0 95 20 25 30 35 40 45 50 51020 50 100 200 500 Minimum mapping quality GIAB benchmark concordance (Phred) Maximum spanned repeat (kb) 94 c Improved mappability of ‘NGS problem exons’ in 193 medically relevant genesNon-gap positions in GRCh37 covered (%) with02 13.5 kb CCS04 reads 060 HG001 ONT Canu + Illumina HG001 PB CLR FALCON HG002 PB CCS Canu (mat) HG001 ONT Canu HG002 PB CLR PBcR HG002 PB CCS Canu (pat) Percentage of problem HiFi reads let you map, and call variants No. of HiFi reads are more concordant with Minimum mapping quality HG002 PB CCS wtdbg2 exons mappable Genes missed by short reads genes GIAB benchmark 100 ABCC6, ABCD1, ACAN, ACSM2B, AKR1C2, ALG1, ANKRD11, BCR, CATSPER2, CD177, CEL, CES1, CFH, CFHR1, CFHR3, CFHR4, CGB, CHEK2, CISD2, CLCNKA, CLCNKB, CORO1A, COX10,c CRYBB2,Improved CSH1, mappabilityCYP11B1, CYP11B2, of ‘NGSCYP21A2, problem CYP2A6, CYP2D6, exons’ CYP2F1, in 193 CYP4A22, medically DDX11, relevant genes with 13.5 kb CCS readsFig. 4 | Impact of read accuracy on de novo assembly. a, The concordance of seven assemblies to the GIAB v.3.3.2 benchmark (Supplementary Table DHRS4L1, DIS3L2, DND1, DPY19L2, DUOX2, ESRRA, F8, FAM120A, FAM205A, FANCD2, FCGR1A, FCGR2A, FCGR3A, FCGR3B, FLG, FLNC, FOXD4, FOXO3, FUT3, GBA, GFRA2, GON4L, GRM5,Percentage GSTM1, GYPA, of GYPB,problem GYPE, HBA1, HBA2, HBG1, HBG2, HP, HS6ST1, IDS, IFT122, IKBKG, 8). Contigs longer than 100­kb were segmented into 100­kbNo. of chunks and aligned to GRCh37. Concordance was measured per chunk and chunks with no IL9R, KIR2DL1, KIR2DL3, KMT2C, KRT17, KRT6A,exons KRT6B, mappable KRT6C, KRT81, KRT86, LEFTY2,Genes LPA, MST1, MUC5B, MYH6, MYH7, NEB, NLGN4X, genes NLGN4Y, NOS2, NOTCH2, NXF5, OPN1LW, OR2T5, OR51A2, PCDH11X, PCDHB4, PGAM1, PHC1, PIK3CA, PKD1, PLA2G10, PLEKHM1, PLG, discordances were assigned concordance of Q51. PB,­PacBio; ONT,­Oxford Nanopore. b, Predicted contiguity of a human assembly based on ability to PMS2, PRB1, PRDM9, PROS1, RAB40AL, RALGAPA1, RANBP2, RHCE, RHD, RHPN2,10 ROCK1,0 ABCC6, SAA1, ABCD1,SDHA, SDHC, ACAN, SFTPA1, ACSM2B, SFTPA2, AKR1C2, SIGLEC14, ALG1, ANKRD11, BCR,152 CATSPER2, CD177, CEL,resolve CES1, repeats CFH, CFHR1, of different CFHR3, CFHR4, lengths CGB, (CHEK2x axis), and percent identities (colored lines)21. The solid line indicates the contiguity of GRCh38. The 97.0% identity SLC6A8, SMG1, SPATA31C1, SPTLC1, SRGAP2, SSX7, STAT5B, STK19, STRC, SULT1A1, SUZ12,CISD2, TBX20, CLCNKA, TCEB3C, CLCNKB, TLR1, TLR6,CORO1A, TMEM231, COX10, TNXB, CRYBB2, CSH1, CYP11B1, CYP11B2, CYP21A2, CYP2A6, CYP2D6, CYP2F1, CYP4A22, DDX11, TRIOBP, TRPA1, TTN, TUBA1A, TUBB2B, UGT1A5, UGT2B15, UGT2B17, UNC93B1, VCY, VWF, WDR72, ZNF419, ZNF592, ZNF674 DHRS4L1, DIS3L2, DND1, DPY19L2, DUOX2, ESRRA, F8, FAM120A, FAM205A, FANCD2, FCGR1A,line is representative FCGR2A, FCGR3A, FCGR3B,of CLR assembliesFLG, FLNC, using standard read-to-read error correction. The points show example CCS and CLR46 assemblies using Canu. [75, 100) ANAPC1, C4A, C4B, CHRNA7, CR1, DUX4, FCGR2B, HYDIN, OTOA, PDPK1, TMLHE FOXD4, FOXO3, FUT3, GBA, GFRA2, GON4L, GRM5, GSTM1, GYPA, GYPB, GYPE, HBA1,Repeat HBA2, HBG1, identity HBG2, andHP, HS6ST1, length IDS, are IFT122, proxies IKBKG for, read accuracy and length. [50, 75) ADAMTSL2, CDY2A, DAZ1, GTF2I, NAIP, OCLN, RPS17 IL9R, KIR2DL1, KIR2DL3, KMT2C, KRT17, KRT6A, KRT6B, KRT6C, KRT81, KRT86, LEFTY2, LPA, MST1, MUC5B, MYH6, MYH7, NEB, NLGN4X, NLGN4Y, NOS2, NOTCH2, NXF5, OPN1LW, OR2T5, OR51A2, PCDH11X, PCDHB4, PGAM1, PHC1, PIK3CA, PKD1, PLA2G10, PLEKHM1, PLG, [25, 50) DAZ2, DAZ3, KIR3DL1, OPN1MW, PPIP5K1 PMS2, PRB1, PRDM9, PROS1, RAB40AL, RALGAPA1, RANBP2, RHCE, RHD, RHPN2, ROCK1, SAA1, SDHA, SDHC, SFTPA1, SFTPA2, SIGLEC14, 11 7 152 (0, 25) NCF1, RBMY1A1 SLC6A8, SMG1, SPATA31C1, SPTLC1, SRGAP2, SSX7, STAT5B, STK19, STRC, SULT1A1, SUZ12, TBX20, TCEB3C, TLR1, TLR6, TMEM231, TNXB, TRIOBP, TRPA1, TTN, TUBA1A, TUBB2B, UGT1A5, UGT2B15, UGT2B17, UNC93B1,5 VCY, VWF, WDR72, ZNF419, ZNF592, ZNF674 0 BPY2, CCL3L1, CCL4L1, CDY1, CFC1, CFC1B, GTF2IRD2, HSFY1, MRC1, OR4F5, PRY, PRY2, SMN1, SMN2, TSPY1, XKRY 16 Coverage requirements for variant calling and de novo assembly. that 2,434 (1,313–2,611; 95% confidence interval) errors in the cur- 2 [75, 100) ANAPC1, C4A, C4B, CHRNA7, CR1, DUX4, FCGR2B, HYDIN, OTOA, PDPK1, TMLHE To evaluate the depth of CCS read coverage required for variant rent GIAB benchmark could be corrected using the CCS reads. [50, 75) ADAMTSL2, CDY2A, DAZ1, GTF2I, NAIP, OCLN, RPS17 Fig. 2 | Mappability of the human genome with CCS reads. a, Percentage of the nongap GRCh37 human genome covered by at least ten reads from 28- calling and assembly, we randomly subsampled from the full data- For structural variants, curator classification was unclear for 11 fold coverage NGS (2 250bp, HiSeq 2500) and CCS (13.5kb) datasets at different [25,mapping 50) DAZ2, quality DAZ3, thresholds. KIR3DL1, OPN1MW, b, Coverage PPIP5K1 of the congenital deafness × set. For SNVs, precision and recall with DeepVariant11 remain7 above of 40 discrepancies, typically because of tandem repeat structure gene STRC in HG002 with 2 250bp NGS reads and 13.5kb CCS reads at a mapping (0,quality 25) NCF1, threshold RBMY1A1 of 10. c, Improvement in mappability with 13.5kb × 99.5% for coverage down to 15-fold; performance decays5 steeply that permits multiple representations of a variant. For the remain- 28 CCS reads for 193 human genes previously reported as medically relevant and problematic0 toBPY2, map CCL3L1, with NGS CCL4L1, reads CDY1,. CFC1, CFC1B, GTF2IRD2, HSFY1, MRC1, OR4F5, PRY, PRY2,below SMN1, 10-fold SMN2, TSPY1, (Supplementary XKRY Fig. 8a). For indels,16 DeepVariant der, 15 of 16 false negative discrepancies were classified as correct in 2 remains comparable to typical NGS performance (>90%) down the benchmark. However, for false positive discrepancies, 11 of 13 Table 1 | Performance of small variant calling with CCSFig. reads2 | Mappability of the human genome with CCS reads. a, Percentage of the nongap GRCh37to 17-fold human coverage genome covered(Supplementary by at least Fig.ten reads8b). fromFor structural28- vari- were classified as errors in the benchmark (Supplementary Fig. 9c,d ants, precision with pbsv is above 95% for all evaluated coverage and Supplementary Table 11). This suggests that the GIAB struc- Platform Variant caller (training foldSNVs coverage NGS (2×250bp, HiSeq 2500)Indels and CCS (13.5kb) datasets at different mapping quality thresholds. b, Coverage of the congenital deafness HiFi reads detect medically-relevant geneslevels. missed Recall by is short above reads90% down to 15-fold coverage and decays tural variant benchmark set is precise but incomplete. model) gene STRC in HG002 with 2 250bpa NGS reads and 13.5kb CCS reads at a mapping quality threshold of 10. c, Improvement in mappability with 13.5kb Precision (%) Recall (%) ×F1 (%) Precision (%) Recall (%) F1 (%) steeply below 10-fold (Supplementary Fig. 8c). For phasing with The high-quality CCS callsets also provide an opportunity to CCS reads for 193 human genes previously reported as medically relevant and problematic to map with NGS reads28. Illumina (NovaSeq) DeepVariant (Illumina model) 99.960 99.940 99.950 99.633 99.413 99.523 WhatsHap, the phase block N50 remains above 150 kb down expand the benchmarks into repetitive and highly polymorphic PacBio (CCS) DeepVariant (CCS model) 99.914 99.959 99.936 96.901 95.980 96.438 to 10-fold coverage (Supplementary Fig. 8d). Mixed-haplotype regions that have been difficult to characterize with confidence PacBio (CCS) DeepVariantWenger, (haplotype- A. M., Tableet99.904 al. 1(2019) | Performance 99.963 Accurate of small99.934 circular variant consensus97.835 calling with long-read97.141 CCS reads 97.486sequencing improveswtdbg2 variantassemblies detection have consistent and assembly size above of2.7 a Gb, human contig N50 using short reads. Adding the CCS DeepVariant callset to the exist- sorted CCSgenome. model) Nature Biotechnology, 37, 1155–1162. around 15 Mb and concordance above Q42 until coverage falls ing GIAB small variant integration pipeline would expand the Illumina (NovaSeq) GATK HaplotypeCaller (no Platform99.852 99.910 Variant99.881 caller99.371 (training 99.156SNVs 99.264 below 15-fold (SupplementaryIndels Fig. 8e-g). benchmark regions by up to 1.3% and 418,875 variants (210,184 filter) model) Precision (%) Recall (%) F1a (%) Precision (%) Recall (%) F1 (%) SNVs and 208,691 indels). For structural variants, only 9,232 of PacBio (CCS) GATK HaplotypeCaller (hard 99.468 99.559 99.513 78.977 81.248 80.097 Revising and expanding GIAB benchmarks. High-quality callsets 18,832 autosomal variant calls overlap benchmark regions, which filter) Illumina (NovaSeq) DeepVariant (Illumina model) 99.960 99.940 from99.950 CCS reads99.633 provide an opportunity99.413 to identify99.523 mistakes in the means that the number of variants in the benchmark would more Precision, recall and F1 of small variant calling measured against the GIAB v.3.3.2PacBio benchmark (CCS) using hap.py. Bold indicates theDeepVariant highest value in each (CCS column. model) Underline indicates99.914 a value higher than the 99.959 GIAB99.936 benchmarks,96.901 particularly95.980 for structural96.438 variants where the than double if all CCS variants calls were incorporated. GATK HaplotypeCaller run on 30-fold Illumina NovaSeq reads. Coverage is 28-fold for PacBio CCS and 30-fold for Illumina NovaSeq. aRows are sorted based on F1 for SNVs. PacBio (CCS) DeepVariant (haplotype- 99.904 99.963 benchmark99.934 is still97.835 in draft form. Sixty97.141 small variant97.486 and 40 structural sorted CCS model) variant discrepancies between the GIAB benchmark (small variant Discussion “HiFi Reads really allow us to call accurate structuralv.3.3.2, variations structural variant and v.0.6) other and types the CCS callsets (DeepVariant We present a protocol for producing highly accurate long reads 21 than deletions (76.90%). The LongRanger callset has Illuminaa precision (NovaSeq) of assemblies GATK (Supplementary HaplotypeCaller Table (no 6). All99.852 assemblies have99.910 haplotype-sorted,99.881 99.371 structural variant99.156 integrated)99.264 were selected for using CCS on the PacBio Sequel System. We apply the protocol 83.79% and recall of 39.83%, again with worse recall for insertions highof contiguityvariationsfilter) with a thatcontig youN50 from can’t 15.43 actually to 28.95 Mb. seeThe with short-read sequencing” (16.41%) than deletions (70.18%). A callset from paftools run on total assembly size is near the expected human genome size for manual curation. Selected variants were spread across variant types, to sequence the human HG002 to 28-fold coverage with an aver- PacBio (CCS) GATK HaplotypeCaller (hard 99.468 99.559 99.513 78.977 81.248 80.097 a linked-read SuperNova assembly has a precision of 64.52% and— JeremyFALCON Schmutz,and wtdbg2. The Faculty Canu assembly Investigator, has a total genome HudsonAlpha discrepancy Institute types of and Biotechnology both inside and outside homopolymers and age read length of 13.5 kb and an average read accuracy of 99.8%. recall of 52.74%. All considered short- and linked-read callsets have size of 3.42 Gb,filter) larger than the expected haploid human genome, tandem repeats. We analyze the CCS reads to call SNVs, indels and structural vari- worse performance than all CCS and CLR callsets in bothPrecision, precision recall and F1because of small variant it resolves calling measuredsome heterozygous against the GIAB alleles v.3.3.2 intobenchmark separate using contigs hap.py. Bold indicates the Forhighest small value in variants,each column. 29Underline of 31 indicates discrepancies a value higher thanin thehomopolymers ants; phase variants into haplotype blocks; and de novo assemble and recall (Supplementary Table 5 and SupplementaryGATK Fig. HaplotypeCaller 5). (Table run on 30-fold2 and IlluminaSupplementary NovaSeq reads. Fig. Coverage 6). is 28-fold for PacBio CCS and 30-fold for Illuminawere NovaSeq. classified aRows are assorted correct based on in F1 forthe SNVs. benchmark. Outside homopoly- the HG002 genome. We demonstrate that CCS reads from a single Short reads from the parents of HG002 were used to identify mers, 19 of 29 were classified as errors in the benchmark. Most of library approach the accuracy of short reads for small variant detec- De novo assembly of CCS reads. Three different algorithms— k-mers unique to one parent and then partition (trio bin) the CCS 39 40 41 42 these benchmark errors (13 of 19) are true variants in L1 elements tion while accessing more of the genome, including medically rel- FALCON , Canu and wtdbg2 —were used to assemble the full reads by haplotype . Three different k-mer sizes were evaluated: called homozygous reference in GIAB (Supplementary Fig. 9a,b and evant genes. CCS reads also enable structural variant detection and CCS read set, which is a mix of paternal and maternal reads. By 21 bp (previously reported for trio binning) and longer k-mers of 21 than deletions (76.90%). The LongRanger callset has a precision of assembliesSupplementary (Supplementary Table 10).Table The 6).identified All assemblies benchmark errorshave overlap de novo assembly at similar contiguity and with markedly higher skipping the initial read-to-read error correction step,83.79% the andalgo -recall51 bp of and 39.83%, 91 bp enabledagain withby the worse accuracy recall of CCS for reads.insertions The 21-mer high contiguity with a contig N50 from 15.43 to 28.95 Mb. The rithms completed 10–100× faster than is typical for long-read binning assigns 35.3% of reads to the mother and 33.6% to the with putative errors in a DeepVariant Illumina whole genome case concordance than noisy long reads. (16.41%) than deletions (70.18%). A callset from paftoolspacb.com/HiFi run on total assemblystudy size (https://github.com/google/deepvariant is near the expected human genome). Of size 745 for putative The CCS performance for SNV and indel calling rivals that of the a linked-read SuperNova assemblyNATURE BI OTECHNOLOGYhas a precision | www.nature.com/naturebiotechnology of 64.52% and FALCON falseand wtdbg2.positive (inThe the Canu callset assembly but not thehas benchmark)a total genome SNVs in the commonly used pairing of BWA and GATK on 30-fold short-read recall of 52.74%. All considered short- and linked-read callsets have size of 3.42case Gb, study, larger 344 than agree the withexpected the CCS haploid callset, human with genome,282 (82.0%) fall- coverage. Interestingly, though the overall accuracy of CCS reads worse performance than all CCS and CLR callsets in both precision because it ingresolves within some large heterozygous interspersed repeats. alleles intoFewer separate of the false contigs negative (in is similar to short reads, direct application of the GATK pipeline and recall (Supplementary Table 5 and Supplementary Fig. 5). (Table 2 andthe Supplementary benchmark but Fig.not the6). callset) SNVs (8%), false negative indels to CCS reads produces inferior results, especially for indels. The Short reads(25%) from and falsethe parentspositive indelsof HG002 (19%) were from used the case to identifystudy agree with major residual error in CCS reads—indels in homopolymers—is De novo assembly of CCS reads. Three different algorithms— k-mers uniquethe CCS to one callset. parent Extrapolating and then partition from manual (trio bin)curation, the CCS we estimate not as frequent in short reads. We suspect that the current GATK, FALCON 39, Canu40 and wtdbg241—were used to assemble the full reads by haplotype42. Three different k-mer sizes were evaluated: CCS read set, which is a mix of paternal and maternal reads. By 21 bp (previously reported for trio binning) and longer k-mers of NATURE BIOTECHNOLOGY | www.nature.com/naturebiotechnology skipping the initial read-to-read error correction step, the algo- 51 bp and 91 bp enabled by the accuracy of CCS reads. The 21-mer rithms completed 10–100× faster than is typical for long-read binning assigns 35.3% of reads to the mother and 33.6% to the

NATURE BIOTECHNOLOGY | www.nature.com/naturebiotechnology PacBio —­ A Trusted Partner With a dedicated global team of scientists, engineers, and bioinformaticians, we are proud to offer end-to-end support to our customers and end users.

Explore How Scientists Use SMRT Sequencing in Their Access Our Training Tools Research, Including More Than 7000 Publications pacb.com/training pacb.com/smrt-resources Review Our Complete Suite of Documentation Learn More About Application-Specific pacb.com/documentation End-To-End Workflows pacb.com/applications Get Support When it Matters – On-Demand Help, Virtually, or Locally pacb.com/support Email: [email protected]

STRUCTURAL WHOLE GENOME ARANT RNA TARGETED COMPLE EPGENETCS Phone: 1.877.920.PACB (7222) ARANT DETECTON SEUENCNG DETECTON SEUENCNG SEUENCNG POPULATONS

“Although we did not have much prior experience on long read sequencing on PacBio platform, every aspect of PacBio operations (starting from “Our Field Application Scientist has always instrument installation, training, been very straightforward, helpful, and library, instrument operation and easy to get on the phone when we needed SMRT Link) has been easy and smooth “right-now support”.” thanks to the amazing PacBio team — Dan New, IBEST Genomics Resources Core that we have.” — Nasun Hah, Director Genomics Sequencing Core, Salk Institute

Sequencing providers around the world are adding the PacBio Systems to their portfolios to meet the demand for highly accurate long-read sequencing. Are you ready to join them? Contact Your Local Sales Team:

North America: [email protected] Europe/Middle East/Africa: [email protected] South America: [email protected] Asia Pacific: [email protected]

For Research Use Only. Not for use in diagnostic procedures. © Copyright 2021, Pacific Biosciences of California, Inc. All rights reserved. Information in this document is subject to change without notice. Pacific Biosciences assumes no responsibility for any errors or omissions in this document. Certain notices, terms, conditions and/or use restrictions may pertain to your use of Pacific Biosciences products and/or third party products. Please refer to the applicable Pacific Biosciences Terms and Conditions of Sale and to the applicable license terms at http://www.pacb.com/ legal-and-trademarks/terms-and-conditions-of-sale/. Pacific Biosciences, the Pacific Biosciences logo, PacBio, SMRT, SMRTbell, Iso-Seq, and Sequel are trademarks of Pacific Biosciences. Femto Pulse and Fragment Analyzer are trademarks of Agilent Technologies Inc. All other trademarks are the sole property of their respective owners. PN: CL100-021921