Applications of next generation sequencing in molecular ecology of non-model organisms

Abstract

As most biologists are probably aware, technological advances in molecular biology during the last few years have opened up possibilities to rapidly generate large-scale sequencing data from non-model organisms at a reasonable cost. In an era when virtually any study organism can 'go genomic', it is worthwhile to review how this may impact molecular ecology. The first studies to put the next generation sequencing (NGS) to the test in ecologically well-characterized species without previous genome information were published in 2007 and the beginning of 2008. Since then several studies have followed in their footsteps, and a large number are undoubtedly under way. This review focuses on how NGS has been, and can be, applied to ecological, population genetic and conservation genetic studies of non-model species, in which there is no (or very limited) genomic resources. Our aim is to draw attention to the various possibilities that are opening up using the new technologies, but we also highlight some of the pitfalls and drawbacks with these methods. We will try to provide a snapshot of the current state of the art for this rapidly advancing and expanding field of research and give some likely directions for future developments.

Figure 1

Scheme showing the workflow from sample to applications of NGS in molecular ecology. We considered three different sources of genetic variation, shown in the central circle of the diagram and indicated by different background shades: genome (light grey), transcriptome (white) and epigenome (grey). Genomic DNA (gDNA), mitochondrial or chloroplast DNA (mtDNA, cpDNA), vectors (for example, BAC), mRNA, non-coding small RNAs or target regions of the genome (targeted DNA) are the samples regarded as starting material (marked with thick circles). Different steps that can be performed before or during sequencing are shown within the grey dotted square representing NGS. gDNA samples, for example, can be used to produce reduced representation libraries (RRLs) and these can be either sequenced directly or used to generate S-RAD markers (sequenced restriction-site associated DNA). Targeted DNA can be generated in several different ways: PCR through amplicon sequencing, microdroplet PCR with RainStorm (RainDance Technologies, Lexington, MA, USA) or through sequence capture (NimbleGen, Roche; SureSelect, Agilent Technologies). mRNA can be used in a deep-serial analysis of gene expression (SAGE) approach or RNA-seq can be performed. gDNA can be also used in epigenetic studies of DNA–protein interactions through ChIP-seq or for studying methylation patterns through BS-seq (high-throughput bisulfite sequencing). The main applications (placed outside of the grey dotted ‘NGS' square) of NGS are gene regulation, expression, transcriptome characterization, development of molecular markers (SNPs, microsatellites, InDels), nucleotide profiling and genome assembly. Each one of these main applications can then be divided further into many interesting areas of molecular ecology (kinship analysis, QTL mapping, and so on). An exceptional case is the use of S-RAD markers that can be used directly in mapping studies without the need of genotyping new developed molecular markers.

Figure 2

cDNA library normalization and ribosomal RNA (rRNA) depletion from total RNA. (a) Agarose gel showing the same double-strand cDNA sample before and after normalization. Normalization decreases the prevalence of highly abundant transcripts (seen as distinctive bands in the un-normalized sample) and equalizes mRNA concentrations in the cDNA library. It will increase the number of genes covered, but unless the sequencing quantity is high, fewer genes will be fully covered. Normalization consequently increases the coverage of most of the sequenced transcript. There is also an increase in the gene discovery rate in a normalized cDNA library, enhancing the identification and analysis of rare transcripts (Cheung et al., 2006). Several normalization methods are reviewed in Bogdanova et al. (2009) but duplex-specific nuclease (DSN) normalization (Zhulidov et al., 2004) has been widely used in recent years. (b) Analysis of RNA with an Agilent 2100 Bioanalyzer (Agilent Technologies) using a pooled RNA sample and the same sample after ribosomal RNA (rRNA) depletion with RiboMinus RNA-seq kit (Invitrogen). Concentrations are showed in fluorescence units (FU) and size in basepairs (nucleotides, nt). Notice the dramatic change of scale of the y axis, which is due to the reduced amount of 18s rRNA in the depleted sample.