1D-Reactor Decentralized MDA for Uniform and ... - ACS Publications


1D-Reactor Decentralized MDA for Uniform and...

0 downloads 81 Views 701KB Size

Subscriber access provided by UNIVERSITY OF CONNECTICUT

Technical Note

1D-reactor Decentralized MDA for Uniform and Accurate Whole Genome Amplification Junji Li, Na Lu, Xulian Shi, Yi Qiao, Liang Chen, Mengqin Duan, Yong Hou, Qinyu Ge, Yuhan Tao, Jing Tu, and Zuhong Lu Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.7b02183 • Publication Date (Web): 30 Aug 2017 Downloaded from http://pubs.acs.org on August 31, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 13

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

1D-reactor Decentralized MDA for Uniform and Accurate Whole Genome Amplification Junji Li a, Na Lu a, Xulian Shi a,b,c, Yi Qiao a, Liang Chen a, Mengqin Duan a, Yong Hou b,c, Qinyu Ge a, Yuhan Tao a, Jing Tu a,*, Zuhong Lu a,* a

State Key Laboratory of Bioelectronics, School of Biological Science and Medical Engineering,

Southeast University, Nanjing, 210096, China b

BGI-Shenzhen, Shenzhen 518083, China

c

China National GeneBank, BGI-Shenzhen, Shenzhen 518120, China

* Correspondence: [email protected], [email protected].

Abstract: Multiple displacement amplification (MDA), a most popular isothermal whole genome amplification (WGA) method, suffers the major hurdle of highly uneven amplification, thus leading to many problems in approaching biological applications related to copy-number assessment. In addition to the optimization of reagents and conditions, complete physical-separation of the entire reaction system into numerous tiny chambers or droplets using microfluidic devices, has been proven efficient to mitigate this amplifying bias in recent works. Here, we present another MDA advance, micro-channel MDA (µcMDA), which decentralizes MDA reagents throughout a one-dimensional slender tube. Due to the double effect from soft partition of high molecular-weight DNA molecules and less-limited diffusion of small particles, µcMDA is shown to be significantly effective at improving the amplification uniformity, which enables us to accurately detect single nucleotide variants (SNVs) with higher efficiency and sensitivity. More importantly, this straightforward method requires neither customized instruments nor complicated operations, making it a ready-to-use technique in almost all biological laboratories.

Keywords: Whole genome amplification; Multiple displacement amplification; Reaction geometry; High-throughput sequencing; Single nucleotide variants; Heterozygosity detection

ACS Paragon Plus Environment

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Over the past 10 years, whole genome amplification (WGA), which is a bridge from low initial genomic DNA to high-throughput sequencing library,1,2 has attracted lots of attention. It is generally acknowledged that WGA methods are divided into two categories: polymerase chain reaction (PCR) based methods3,4 and multiple displacement amplification (MDA) based methods.5-7 To guarantee high physical coverage of genome and good uniformity of amplicons, non-specific amplification is desired, especially in the early stage of amplifying.8 Due to the preference of primers, PCR products cover fewer regions of genome but always perform better in maintaining the uniformity of these amplified regions.9 The isothermal methods based on MDA can cover most of the genome, but are challenged by the uneven amplification, which results in the intra-genome non-uniformity and low resolution for heterozygote detection between genomes.10 Hybrid methods, such as MALBAC11,12 and PicoPLEX, use random priming and poikilothermic pre-amplification in the early stage, followed by PCR amplification with limited circles. Coverage and uniformity are both intermediately improved due to the quasi-linear pre-amplicons and limited PCR cycles,13 but replacing phi29 polymerase of high fidelity by Bst polymerase introduces more bias and complicates the final products.8 Among these aforementioned methods, MDA is most frequently adopted because of its simple operation, reliable proofreading and high amplifying efficiency. Microfluidic devices provide powerful and flexible platforms for WGA and extend its application.14,15 Due to its isothermal property, MDA is more compatible than PCR to be integrated into a microfluidic system and has been widely employed. One of the most significant applications is single cell sequencing,16-18 which has been introduced to studies of environmental microbial diversity, genetic alterations in cancer cells and germline gene transfer.13,19 Generally, these applications only incorporate MDA directly with microfluidic systems. The uneven amplification still restricts MDA from being efficiently applied in cases associated with copy number analysis.20 Physical segregation of analyte molecules using microfluidic devices21 brings the dawn for improving the uniformity of MDA. By separating reagents directly into nanoliter-scale chambers22 or wells23, amplification bias is greatly reduced and more heterozygous genomic diversities are detected. Later on, with the widespread of droplet-based microfluidic techniques, several droplet-MDA protocols emerged almost simultaneously.24-26 These techniques further reduced the volume of reactors to picoliter scale, making them markedly effective against contaminations and amplification bias.27 The performance of MDA is significantly improved thanks to the dual effect of increasing parallel-assay numbers and limiting the influence of adverse reactions from contaminations and amplification-preferred templates. Recently, droplet-based MDA has been applied in a metagenome study,28 indicating that quantitative relationships in metagenomes from low biomass environments can be well recovered. Microfluidic techniques are usually considered to be convenient in use. However, chip design and fabrication, which is quite time-consuming to be outsourced, requires totally different skills and equipment for biological and chemical researchers.29 Alternatively, commercialized droplet generators are relatively expensive. In this technical note, we present a novel MDA method, micro-channel MDA (µcMDA), which significantly improves the uniformity of amplification. In this method, MDA is improved by only a few simple operations with no extra experimental appliance and specifically-customized accessories needed. Compared with conventional MDA, we demonstrate that µcMDA which is conducted in a micro channel with small inner diameter is able to efficiently suppress amplification bias.

ACS Paragon Plus Environment

Page 2 of 13

Page 3 of 13

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

EXPERIMENTAL SECTION All the glass and plastic consumables were properly cleaned, dried and UV-treated before use. Purified genomic DNA of YH-1 cells (immortalized cells of a Han Chinese individual, provided by BGI Shenzhen) was extracted (QIAamp DNA Mini Kit, Qiagen), then quantitated using commercially available kits (Qubit dsDNA HS Assay Kit, ThermoFisher Scientific). Commercialized MDA kit (REPLI-g MIDI kit, Qiagen) was used according to the manufacturer’s protocol, with a few modifications. In both µcMDA and MDA assays, 10 ng genomic DNA was added into the reaction buffer for a total 50 microliter volume. After alkaline denaturation and neutralization processes, polymerases were separately added into the reaction tubes. Afterwards, conventional MDA tube was incubated in a thermo cycler at 30°C for 16 h, followed by 3 min’s inactivation at 65°C. The reaction mixture of µcMDA was gently aspirated into a polytetrafluoroethylene (PTFE) capillary tubing (320 micron inner diameter, Adtech Polymer Engineering Ltd.). This tubing was pre-treated with 0.1% w/w bull serum albumin (BSA) for 1 h. Then this tubing was coiled around a metal cord-winder, and both ends were crimped with a sealing clip. All the preparations were performed at 4°C to prevent the amplification from starting in advance. Eventually, the tubing-coiling winder was immersed in a 30°C water bath for 16 h, and then moved into another water bath at 65°C for 5 min (Figure 1). Three parallel µcMDA assays were conducted to validate the effectiveness and reproducibility of our method. Amplification products in both centrifuge tube and capillary tubing were respectively collected, purified and constructed into Illumina libraries. Each library was sequenced on Illumina HiSeq4000 using 2 × 150 paired-end reads. Qualified sequencing data was aligned to the hg19 reference genome using the BWA short sequence alignment software,30 and then statistically analysed through QualiMap2.31 single nucleotide variants (SNV) were called by using SAMtools32 and Varscan2.33 Further results were calculated and exhibited by Perl scripts. RESULTS AND DISCUSSION High efficiency of amplification. Compared with conventional in-tube assay, µcMDA produced products of comparable amounts (Table S-1) after the same period of time, which indicates that µcMDA is able to maintain the high amplifying efficiency of MDA. It is inferred that a certain degree of geometry alteration of reaction space from cube-like to linear would not affect the holistic process of reaction. The diffusion of ions and small molecules could alleviate the drastic change in concentrations during the reaction process. Therefore, instead of complete physical isolation, µcMDA is able to maintain the connectivity of the whole reaction system and preserve the high efficiency of amplification. Better coverage breadth and genome recovery. After analysing the 4 × low-coverage whole genome sequencing (LWGS) data,10 it was found that the parallel µcMDA assays revealed similar performance (Table S-2). Thus, we chose the sample #1 data, which showed an intermediate performance, to conduct further deep-sequenced data analysis for the subsequent investigation of amplification bias and variation-detection ability. From the deep-sequenced data, it was shown that µcMDA reads covered wider areas in genome than MDA reads with same sequencing depth (Table S-3). This advantage became more evident when comparing the coverage of genome with high sequencing depth. Then genomic areas of at least 20 × sequencing depth were checked, showing that µcMDA reads covered 13.28% more areas. The result supported that smaller

ACS Paragon Plus Environment

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

quantities of µcMDA products were able to recover the target genome. By 13 × sequencing data, products of µcMDA recovered over 95% of the genome, while products of MDA required 25 × (Figure S-1), about twice the consumption, to achieve the same performance. Improved amplification uniformity. To evaluate the genomic coverage distribution, we separated the whole genome of hg19 into bins of fixed 40 Kb size, then calculated the standardized average depth of each bin (details provided in Supporting information). Fixed-size binning was used because it is able to show more primitive results than those optimized binning strategies. Read depth of each bin was normalized by unamplified bulk data to exclude sample specificity. The coverage distribution of µcMDA reads was significantly more even than that of MDA reads in the whole genome (Figure S-2) with the coefficient of variation (CV) being 0.37 (µcMDA) and 0.78 (MDA) respectively at this bin size. From the histogram of the read depth over the entire chromosome X (Figure 2a), it was observed that all the bins of the µcMDA sequencing data were generally even in depth with CV of 0.37, while numerous abnormal bins were observed in the sequencing data of MDA (CV = 0.83). The uniformity superiority was still prominent even if we zoomed in and narrowed the bin size to 2 Kb, (Figure 2b). Lorenz curves were also plotted to validate the coverage uniformity of genome (Figure 3a). Central diagonal line represented theoretical perfect uniformity of read distribution. By comparing the Lorenz curves of bulk, µcMDA and MDA samples, it was found that µcMDA sample showed closer performance to the unamplified bulk sample, while the conventional MDA sample deviated obviously from the bulk and µcMDA samples. All the curves mentioned above were obtained from ~30× mean sequencing-depth data. There was no difference of amplification conditions between µcMDA and MDA except for the reactors used. Therefore, the geometry change of the reactor seemed to be the major cause of the uniformity improvement of µcMDA. In order to investigate the mechanism amplification bias, the auto-correlations of base-level coverage in chromosome 1 were calculated at various depths to examine coverage correlations at all length scales (Figure S-3).34 We identified the characteristic correlation length of ~7 Kb, which was independent with sequencing depth. This amplicon-level parameter reflected the intrinsic non-uniformity of amplification products, which can eliminate the influences of the steps in sequencing library construction. From the correlation variation, it was found that the improvement of µcMDA mainly resulted from the inter-amplicon level and could be better if longer DNA templates were input. Besides, µcMDA experiments using 600 pg genomic DNA (~100 cells) as initiation were also conducted to evaluate its performance under low-input condition. As is shown in Figure S-4, the improvement was still remarkable. Correlation with reaction geometry. In order to quantitatively compare the influence of reactors with different geometries, a simulation of the change in average distance between the central unit and the other units was introduced, where the distance was relative to the change of inner diameter of capillary tubing (Supporting information). In this simulation, reactors in different geometries were divided into small reaction units according to the template concentration. The central unit of any cylindrical reactors was mostly subject to be influenced by other units. From the simulation result, significantly sharp increase of average distance was observed as the inner diameter of the reactor decreases (Figure 3b). Here we chose inner diameter as the variable, which is a more appropriate parameter than internal surface area for regular-shaped reactors, to simplify the calculation and construction of diffusion model. The increase in average distance revealed that most reaction units in the tubing were remarkably separated with each other. Therefore, all of the

ACS Paragon Plus Environment

Page 4 of 13

Page 5 of 13

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

units were soft-partitioned and the local depletion of big molecules,35 such as primers, was not able to be replenished timely and abundantly by limited supply from the few units around, let alone the units far apart. In our study, a reaction system of 50 microliters formed a liquid column of 0.62 meter long in the thin channel with an average distance of 0.17659 meter, which is about 100 times of the average distance in a cylinder whose diameter equals its length to approximate the experimental condition in micro centrifuge tube. Therefore, most of the remaining units are more distant from the central unit in micro channel than in tube. As a result, over-amplification36 caused by the snowballing effect of random priming would be confined in a local range by using the quasi-1D tubing, thus ensuring the high overall amplification uniformity. µcMDA experiments were also performed in tubings of different inner diameters to validate the relationship between amplification uniformity and reaction geometry (Figure S-5). It was observed that the CV value became larger as the inner diameter increased. Accurate SNV detection with higher efficiency and sensitivity. In terms of SNV detection (Supporting information), µcMDA gives better detection rate for both homozygous and heterozygous SNVs on chromosome 1 of diploid YH-1 cell (Table 1). 83.42% high-confidence SNVs were identified using µcMDA, in contrast to 67.40% by conventional MDA with the same data size. Then the allelic dropout (ADO) rate was examined by counting the loss-of-heterozygosity events from the comparison of SNVs in µcMDA/MDA and in bulk. Due to the relatively high initial quantity, the ADO rate of µcMDA was as low as 3.12%, better than 3.35% of MDA. The improvement of both detection efficiency and ADO rate resulted from the better amplifying uniformity of µcMDA. Furthermore, the false positives and errors were counted by comparing SNVs in µcMDA/MDA with SNVs in bulk. Both false positive rate and error rate of µcMDA were in the same magnitude but superior to those of MDA. This indicated that geometric change of reactor in µcMDA would not affect the accuracy derived from phi29 polymerase. Subsequently, we reduced the input data to 10 ×. It turned out that µcMDA still maintained relatively higher detection rate for SNVs, which means that µcMDA was more sensitive than conventional MDA in detecting SNVs with low sequencing depth (Table S-4). Low requirement for equipment and operation. By using µcMDA, amplification uniformity is highly improved without any decrease of amplifying efficiency, while almost no extra instrument but a few commercially available accessories are required in µcMDA experiment. Although higher amplification uniformity can also be achieved by the complete isolation of reaction units using micro-reactor MDA22,23 or emulsion MDA24-26 protocols. However, these approaches require either complicated operations or considerable additional cost (Table 2). The simple experimental procedures and operations enable µcMDA to be more appropriate for tracing samples with less unexpected loss. Besides, abandoning emulsions can also avoid the potential interference from high concentration of surfactants and heterogeneous reaction system. Conclusions We proposed an improved MDA technique, µcMDA, which accommodates conventional MDA reaction in a quasi-1D micro channel. By comparing with conventional MDA products under the same reaction conditions, we demonstrated that our approach can dramatically enhance the amplifying uniformity throughout the entire genome by suppressing the over-amplification in a local range. In terms of SNV detection, µcMDA exhibits higher efficiency and sensitivity for both homozygous and heterozygous SNVs while maintaining the same high-level accuracy as

ACS Paragon Plus Environment

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

conventional MDA. Compared to previous techniques like micro-reactor MDA and emulsion MDA, our method is easy and convenient in practical operations with low demand for customized equipment or accessories. It is believed that this novel and promising method could provide a more convenient approach to access the genomic information, especially for researches requiring quantitative analysis of higher resolution. Acknowledgements This work was supported by the National Key Project of China (No. 2016YFA0501600) and project 61571121 of National Natural Science Foundation of China. Supporting information The Supporting Information is available free of charge on the ACS Publications website at DOI: Amplification efficacies by Qubit 2.0 quantitation, summaries of LWGS data from parallel µcMDA assays, summarized analysis results of whole genome deep-sequenced data, genome recovery of raw sequence data, standardized fixed-size binning method, comparison of sequencing coverage on all chromosomes, base-level auto-correlation at various depths, CV-binsize figure of low initial template concentration data, simulation method of average central distance, amplification uniformity comparison of µcMDA experiments with different inner diameter, SNV analysis method, SNV-detecting sensitivity of lower input data (PDF) References (1) Shapiro, E.; Biezuner, T.; Linnarsson, S. Nat. Rev. Genet. 2013, 14, 618-630. (2) Zhang, L.; Cui, X. F.; Schmitt, K.; Hubert, R.; Navidi, W.; Arnheim, N. Proc. Natl. Acad. Sci. U. S. A. 1992, 89, 5847-5851. (3) Telenius, H.; Carter, N. P.; Bebb, C. E.; Nordenskjold, M.; Ponder, B. A. J.; Tunnacliffe, A. Genomics 1992, 13, 718-725. (4) Lao, K.; Xu, N. L.; Straus, N. A. Biotechnol. J. 2008, 3, 378-382. (5) Dean, F. B.; Nelson, J. R.; Giesler, T. L.; Lasken, R. S. Genome Res. 2001, 11, 1095-1099. (6) Dean, F. B.; Hosono, S.; Fang, L. H.; Wu, X. H.; Faruqi, A. F.; Bray-Ward, P.; Sun, Z. Y.; Zong, Q. L.; Du, Y. F.; Du, J.; Driscoll, M.; Song, W. M.; Kingsmore, S. F.; Egholm, M.; Lasken, R. S. Proc. Natl. Acad. Sci. U. S. A. 2002, 99, 5261-5266. (7) Lasken, R. S. Biochem. Soc. Trans. 2009, 37, 450-453. (8) Chen, M.; Song, P.; Zou, D.; Hu, X.; Zhao, S.; Gao, S.; Ling, F. PloS one 2014, 9, e114520. (9) Huang, L.; Ma, F.; Chapman, A.; Lu, S.; Xie, X. S. Annu. Rev. Genomics Hum. Genet. 2015, 16, 79-102. (10) Hou, Y.; Wu, K.; Shi, X.; Li, F.; Song, L.; Wu, H.; Dean, M.; Li, G.; Tsang, S.; Jiang, R.; Zhang, X.; Li, B.; Liu, G.; Bedekar, N.; Lu, N.; Xie, G.; Liang, H.; Chang, L.; Wang, T.; Chen, J., et al. Gigascience 2015, 4, 37. (11) Lu, S.; Zong, C.; Fan, W.; Yang, M.; Li, J.; Chapman, A. R.; Zhu, P.; Hu, X.; Xu, L.; Yan, L.; Bai, F.; Qiao, J.; Tang, F.; Li, R.; Xie, X. S. Science 2012, 338, 1627-1630. (12) Zong, C.; Lu, S.; Chapman, A. R.; Xie, X. S. Science 2012, 338, 1622-1626. (13) Gawad, C.; Koh, W.; Quake, S. R. Nat. Rev. Genet. 2016, 17, 175-188. (14) Yoshino, T.; Tanaka, T.; Nakamura, S.; Negishi, R.; Hosokawa, M.; Matsunaga, T. Anal. Chem. 2016, 88, 7230-7237.

ACS Paragon Plus Environment

Page 6 of 13

Page 7 of 13

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

(15) Yu, Z. L.; Lu, S. J.; Huang, Y. Y. Anal. Chem. 2014, 86, 9386-9390. (16) Fritzsch, F. S. O.; Dusny, C.; Frick, O.; Schmid, A. Annu. Rev. Chem. Biomol. Eng. 2012, 3, 129-155. (17) Sanchez, J. L. A.; Joda, H.; Henry, O. Y. F.; Solnestam, B. W.; Kvastad, L.; Akan, P. S.; Lundeberg, J.; Laddach, N.; Ramakrishnan, D.; Riley, I.; Schwind, C.; Latta, D.; O'Sullivan, C. K. Anal. Chem. 2017, 89, 3378-3385. (18) Fu, Y. S.; Chen, H.; Liu, L.; Huang, Y. Y. Anal. Chem. 2016, 88, 10795-10799. (19) Wang, Y.; Navin, N. E. Molecular Cell 2015, 58, 598-609. (20) Navin, N. E. Genome Biol. 2014, 15, 452. (21) Eastburn, D. J.; Sciambi, A.; Abate, A. R. Anal. Chem. 2013, 85, 8016-8021. (22) Marcy, Y.; Ishoey, T.; Lasken, R. S.; Stockwell, T. B.; Walenz, B. P.; Halpern, A. L.; Beeson, K. Y.; Goldberg, S. M.; Quake, S. R. PLoS Genet. 2007, 3, 1702-1708. (23) Gole, J.; Gore, A.; Richards, A.; Chiu, Y. J.; Fung, H. L.; Bushman, D.; Chiang, H. I.; Chun, J.; Lo, Y. H.; Zhang, K. Nat. Biotechnol. 2013, 31, 1126-1132. (24) Fu, Y.; Li, C.; Lu, S.; Zhou, W.; Tang, F.; Xie, X. S.; Huang, Y. Proc. Natl. Acad. Sci. U. S. A. 2015, 112, 11923-11928. (25) Nishikawa, Y.; Hosokawa, M.; Maruyama, T.; Yamagishi, K.; Mori, T.; Takeyama, H. PloS one 2015, 10, e0138733. (26) Sidore, A. M.; Lan, F.; Lim, S. W.; Abate, A. R. Nucleic Acids Res. 2016, 44, e66. (27) Tay, A.; Kulkarni, R. P.; Karimi, A.; Di Carlo, D. Lab Chip 2015, 15, 4379-4382. (28) Hammond, M.; Homa, F.; Andersson-Svahn, H.; Ettema, T. J.; Joensson, H. N. Microbiome 2016, 4, 52. (29) Chen, Z.; Fu, Y.; Zhang, F.; Liu, L.; Zhang, N.; Zhou, D.; Yang, J.; Pang, Y.; Huang, Y. Lab Chip 2016, 16, 4512-4516. (30) Li, H.; Durbin, R. Bioinformatics 2009, 25, 1754-1760. (31) Okonechnikov, K.; Conesa, A.; Garcia-Alcalde, F. Bioinformatics 2016, 32, 292-294. (32) Li, H.; Handsaker, B.; Wysoker, A.; Fennell, T.; Ruan, J.; Homer, N.; Marth, G.; Abecasis, G.; Durbin, R.; Genome Project Data Processing, S. Bioinformatics 2009, 25, 2078-2079. (33) Koboldt, D. C.; Zhang, Q.; Larson, D. E.; Dong, S.; Mclellan, M. D.; Ling, L.; Miller, C. A.; Mardis, E. R.; Li, D.; Wilson, R. K. Genome Res. 2012, 22, 568. (34) Zhang, C.-Z.; Adalsteinsson, V. A.; Francis, J.; Cornils, H.; Jung, J.; Maire, C.; Ligon, K. L.; Meyerson, M.; Love, J. C. Nat. Commun. 2015, 6. (35) Welch, T. W.; Corbett, A. H.; Thorp, H. H. J. Phys. Chem. 1995, 99, 11757-11763.

(36) de Bourcy, C. F. A.; De Vlaminck, I.; Kanbar, J. N.; Wang, J.; Gawad, C.; Quake, S. R. PloS one 2014, 9, e105585.

ACS Paragon Plus Environment

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 1. Schematic illustration of µcMDA experimental process. In conventional MDA (red), reagents were distributed in three-dimensional space, making preferred templates over-amplified by timely reactant supplement. In µcMDA (blue), reagents were uniformly distributed in the narrow channel. The quasi-1-dimentional geometry strongly restricted diffusion of large molecules such as enzymes, templates and pre-amplicons, thus cutting off the positive feedback of over-amplification and resulting in even-amplification.

ACS Paragon Plus Environment

Page 8 of 13

Page 9 of 13

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Figure 2. Comparison of sequencing coverage between µcMDA and MDA reads. (a) Read depth distribution over chromosome X of YH-1 cell line. The entire chromosome was divided into 40 Kb bins, and the standardized mean coverage depth of µcMDA reads (blue bars, above X axis) as well as the MDA reads (red bars, below X axis) in each bin were calculated. (b) Zoom in on the 70 Mb to 78 Mb region of chromosome X (dashed box with gray background in (a)). Bin size was adjusted to 2 Kb to ensure the same sampling rate with (a).

ACS Paragon Plus Environment

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 3. Analysis of read distribution uniformity. (a) Lorenz curves of bulk, µcMDA and MDA samples depict the read coverage uniformity across whole genome. The diagonal line which represents perfectly uniform coverage, is the reference line of the curves calculated from unamplified bulk (green), µcMDA (blue) and conventional MDA (red) reads with ~30× depth. (b) Simulation of the average distance between central reaction unit with other units. Reaction space of different geometry was divided into minimized units. Red triangle and rhombus on the curve respectively represent situations of µcMDA and MDA reactions. The gray dash shows the internal surface area of micro-channel reactors with different inner diameters.

ACS Paragon Plus Environment

Page 10 of 13

Page 11 of 13

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Table 1. Summary of the comparison between different methods for SNV detection on chromosome 1 of normal diploid YH-1 cell Sample type* Parameter

µcMDA

MDA

Bulk

Total SNVs

236464

191056

283476

Detection rate

83.42%

67.40%

N/A

Heterozygous SNVs

99254

79123

117367

Detection rate

84.57%

67.42%

N/A

Homozygous SNVs

137210

111933

166109

Detection rate

82.60%

67.39%

N/A

ADO rate

3.12%

3.35%

N/A

SNV error rate

0.05%

0.13%

N/A

False-positive rate

7.50%

8.13%

N/A

* Calculation was based on sequencing data of larger than 30 × data size.

ACS Paragon Plus Environment

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 13

Table 2. Comparison of µcMDA with reported MDA methods. MDA

Micro-reactor MDA**

Emulsion MDA

µcMDA

Uniformity

low

medium

high

high

Efficiency

high

medium

high

high

Equipment

N/A

~$0-8000

~$1000

N/A

Mould

N/A

~$100-2000

~$100

N/A

Amplifying performance

Additional setup costs*

Additional process costs and hands-on times*

time

cost

time

cost

time

cost

time

cost

Device fabrication

N/A

N/A

3h-7h

~$10-50

3h

$20

N/A

N/A

System adjusting

N/A

N/A

0-1h

N/A

30min

N/A

N/A

N/A

Experimental process

N/A

N/A

5min-1h

N/A

25min

$1

15min

$0.5

* Costs are estimated according to the lowest quoted prices in China and then converted into dollars. As lithography instrument is not commonly equipped in biological laboratories, microfluidic-involved methods take the unified processing mode of outsourced fabrication of moulds and on-site fabrication of microfluidic devices. N/A represents not applicable. ** Differences between micro-reactor MDA methods are large. Typically, High degree of automation requires more complex and expensive chip fabrication and control but significantly reduces operating time.

ACS Paragon Plus Environment

Page 13 of 13

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

for TOC only

ACS Paragon Plus Environment