Sequence depth, not PCR replication, improves ecological inference from next generation DNA sequencing.

Smith DP, Peay KG. · 2014

Read Sequence depth, not PCR replication, improves ecological inference from next generation DNA sequencing. by Smith DP, Peay KG. (2014) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc3938664.txt.

Introduction

Dylan P. Smith and Kabir G. Peay ask a budget question for DNA-based ecology. Their paper appeared in PLoS ONE in 2014. Is it better to repeat the DNA copying step? Or is it better to sequence each sample more deeply? They worked with soil fungi from pine forests in North America. Three soil samples were sequenced on two platforms. The team varied how many copy rounds were pooled each time.

Pooling more copy rounds changed little. Measures of diversity within and between samples stayed put. Deeper sequencing helped a lot. It cut the false gaps seen when one sample was read twice. The newer platform also found far more species. Diversity patterns from the two platforms still agreed well. The authors advise labs to spend on depth, not on copy repeats.

Catalog record

CREATORSmith DP, Peay KG.
YEAR2014
COLLECTIONPAPERS
FORMATTXT
FILE SIZE50 KB
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC3938664

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Sequence Depth, Not PCR Replication, Improves Ecological Inference from Next Generation DNA Sequencing By Dylan P Smith, Kabir G Peay, Christina A Kellogg Published in: PLoS ONE DOI: 10.1371/journal.pone.0090234 ABSTRACT Abstract Recent advances in molecular approaches and DNA sequencing have greatly progressed the field of ecology and allowed for the study of complex communities in unprecedented detail. Next generation sequencing (NGS) can reveal powerful insights into the diversity, composition, and dynamics of cryptic organisms, but results may be sensitive to a number of technical factors, including molecular practices used to generate amplicons, sequencing technology, and data processing. Despite the popularity of some techniques over others, explicit tests of the relative benefits they convey in molecular ecology studies remain scarce. Here we tested the effects of PCR replication, sequencing depth, and sequencing platform on ecological inference drawn from environmental samples of soil fungi. We sequenced replicates of three soil samples taken from pine biomes in North America represented by pools of either one, two, four, eight, or sixteen PCR replicates with both 454 pyrosequencing and Illumina MiSeq. Increasing the number of pooled PCR replicates had no detectable effect on measures of α- and β-diversity. Pseudo-β-diversity – which we define as dissimilarity between re-sequenced replicates of the same sample – decreased markedly with increasing sampling depth. The total richness recovered with Illumina was significantly higher than with 454, but measures of α- and β-diversity between a larger set of fungal samples sequenced on both platforms were highly correlated. Our results suggest that molecular ecology studies will benefit more from investing in robust sequencing technologies than from replicating PCRs. This study also demonstrates the potential for continuous integration of older datasets with newer technology. INTRODUCTION Next generation DNA sequencing (NGS) has changed the face of microbial ecology in the space of a few years. As a result, we have gained unprecedented insight into the community dynamics of morphologically cryptic organisms such as fungi, bacteria and viruses [1] [2] [3]. However, the outcome of NGS based ecological inquiry may be sensitive to technical practices that in many cases have not been adequately tested. For instance, the assumption that NGS read counts accurately reflect absolute abundance in ecological analyses may not be appropriate due to taxon specific PCR and sequencing biases [4]. These technical practices have important effects on our view of underlying biological reality, but also on the allocation of resources (time, money, reagents) that often define the scope of ecological inquiry. Early optimization of NGS methods has focused on correcting platform specific sequencing issues, such as the known homopolymer error rates in 454 pyrosequencing [5], often with bioinformatic solutions. However, potential distortions may also arise prior to DNA sequencing during sample collection [6], DNA extraction [7] [8], or PCR amplification [9] [10]. Recognition of these problems has led to a loosely knit collection of best lab and bioinformatics practices that have emerged in the microbial ecology literature and that are aimed at increasing the robustness of whole community amplification. Among other things, such practices include the use of hot-start Taq polymerase, reducing the number of amplification cycles, and the pooling of multiple PCR replicates per sample [11]. The necessity of pooling PCR replicates is thought to arise from stochasticity in the PCR process that results in variable composition of DNA fragments across individual PCR reactions. Possible causes for this may be sampling effects that lead to variation in the initial population of DNA template used to start the reaction, slight variation in initial conditions, or priority effects of amplification in the early rounds of PCR. The few studies that have actually reported results from replicate NGS of the same sample (e.g. [8] [12]) have found sample-to-sample variance in sequence composition that seem to support the importance of stochastic PCR effects. However, these studies have focused on the comparison of individual PCR replicates that were sequenced separately, and therefore the extent to which their conclusions rely on PCR or sequencing stochasticity is still unknown. Though many studies have suggested pooling PCR replicates prior to sequencing, to our knowledge no study has directly tested whether samples comprised of multiple, pooled PCR replicates capture a more robust sample of the true diversity within the sample. Because there have been no explicit tests of PCR pooling, papers vary wildly in the number of pooled PCR replicates they use or recommend. While three appears to be a somewhat canonical number (e.g. it has been adopted by the Earth Microbiome Project [13]) other studies have used a single PCR replicate [14], five [15] or even ten [16]. The rapid rise and fall of NGS sequencing platforms also raises major concerns about portability of data across studies. One of the major advantages of DNA based community profiling is the collection of standardized data that can be compared or combined across studies. However, if the observed structure of a community is platform dependent it would seriously weaken the additive nature of sequencing efforts in microbial ecology [17]. In this study our primary goals were (1) to determine quantitatively the number of pooled PCR replicates that maximizes ecological inference in NGS studies and (2) to test the robustness of ecological inferences about community structure using two different NGS sequencing platforms. We did this using a two-pronged experimental approach where the same DNA sample was sequenced from a pool of 1, 2, 4, 8 or 16 separate PCR reactions using both Roche's 454 Pyrosequencing (454) and Illumina's MiSeq (MiSeq). We then compared patterns of α- and β-diversity (the primary response variables in most community ecology studies) among samples and replicates. In addition, we used both NGS platforms to sequence a larger set of soil samples taken from Pine forests in geographically distinct parts of North America where we expected to see differences in community composition. Based on previous studies, we hypothesized that increasing the number of PCR replicates prior to sequencing would increase α-diversity, reduce β-diversity and increase reproducibility by averaging out PCR noise. Surprisingly, we found that increasing PCR replication did not meaningfully change any of our ecological response variables and may be a poor investment of resources in molecular ecology studies. By contrast, increased sequencing depth markedly improved estimates of β-diversity using both NGS platforms. In addition, we found that community sequencing results from 454 and MiSeq provide largely similar results, suggesting that data from the two platforms can be combined in a meaningful way. MATERIALS AND METHODS Experimental design We investigated the effects of PCR replication and sequencing platforms on ecological inference using samples from an ongoing project to characterize ectomycorrhizal fungal communities across North America pine forests. To test the hypothesis that pooling PCR replicates improves ecological inference, we selected three soil samples taken from two sites in Oregon (42.79° E –121.62° N) and one in Connecticut (41.82° E, –72.96° N). To test whether NGS sequencing platform affects ecological inference, we selected 60 samples from two sites in North Carolina (NC1 36.01° E, –78.97° N, NC2 35.99° E, –79.10° N), two sites in California (CA1 37.84° E, –119.94° N, CA2 37.81° E, –119.91° N), and two sites in Alaska (AK1 64.77° E, –148.27° N, AK2 64.76° E, –148.25° N) (for all sites n = 10). All sample sites in

Read the complete text