While metagenomics focuses on studying the genomic content and on identifying which microbes are present within a community, metatranscriptomics can be used to study the diversity of the active genes within such community, to quantify their expression levels and to monitor how these levels change in different conditions (e.g., physiological vs. pathological conditions in an organism). The advantage of metatranscriptomics is that it can provide information about differences in the active functions of microbial communities which appear to be the same in terms of microbe composition.
The microbiome has been defined as a microbial community occupying a well-defined habitat. They are ubiquitous and extremely relevant for the maintenance of” the characteristic of the environment in which they reside and an imbalance in these communities can affect negatively the activity of the setting in which they reside. To study these communities, and to then determine their impact and correlation with their niche, different omics- approaches have been used. While metagenomics allows to obtain a taxonomic profile of the sample, metatrascriptomics provides a functional profile by analysing which genes are expressed by the community. It is possible to infer what genes are expressed under specific conditions, and this can be done using functional annotations of expressed genes.
Since metatranscriptomics focuses on what genes are expressed, it allows to understand the active functional profile of the entire microbial community. The overview of the gene expression in a given sample is obtained by capturing the total mRNA of the microbiome and by performing a whole metatranscriptomics shotgun sequencing.
Tools and techniques
Although microarrays can be exploited to determine the gene expression profiles of some model organisms, NGS is the preferred technique in metatranscriptomics. The protocol that is used to perform a metatranscriptome analysis may vary depending on the type of sample that needs to be analysed. Indeed, many different protocols have been developed for studying the metatranscriptome of microbial samples. Generally, the steps include sample harvesting, RNA extraction (different extraction methods for different kinds of samples have been reported in the literature), mRNA enrichment, cDNA synthesis and preparation of metatranscriptomic libraries, sequencing and data processing and analysis. mRNA enrichment is one of the trickiest parts. Different strategies have been proposed:
- removing rRNA through Ribosomal RNA capture
- using a 5-3 exonuclease to degrade processed RNAs (mostly rRNA and tRNA
- adding poly(A) to mRNAs by using a polyA polymerase (in E. coli)
- using antibodies to capture mRNAs that bind to specific proteins
The last two strategies are not recommended as they have been reported to be highly biased. 
A typical metatranscriptom¬e analysis pipeline:
- maps reads to a reference genome or
- performs de novo assembly of the reads into transcript contigs and supercontigs
The first strategy maps reads to reference genomes in databases, to collect information that is useful to deduce the relative expression of the single genes. Metatran¬scriptomic reads are mapped against databases using alignment tools, such as Bowtie2, BWA, and BLAST. Then, the results are annotated using resources, such as GO, KEGG, COG, and Swiss-Prot. The final analysis of the results is carried out depending on the aim of the study. One of the latest metatranscriptomics techniques is stable isotope probing (SIP), which has been used to retrieve specific targeted transcriptomes of aerobic microbes in lake sediment. The limitation of this strategy is its reliance on the information of reference genomes in databases. The second strategy retrieves the abundance in the expression of the different genes by assembling metatranscrip¬tomic reads into longer fragments called contigs using different softwares. So, its limits depend on the software that is used for the assembly. The Trinity software for RNA-seq, in comparison with other de novo transcriptome assemblers, was reported to recover more full-length transcripts over a broad range of expression levels, with a sensitivity similar to methods that rely on genome alignments. This is particularly important in the absence of a reference genome. A quantitative pipeline for transcriptomic analysis was developed by Li and Dewey  and called RSEM (RNA-Seq by Expectation Maximization). It can work as stand-alone software or as a plug-in for Trinity. RSEM starts with a reference transcriptome or assembly along with RNA-Seq reads generated from the sample and calculates normalized transcript abundance (meaning the number of RNA-Seq reads cor-responding to each reference transcriptome or assembly). Although both Trinity and RSEM were designed for tran¬scriptomic datasets (i.e., obtained from a single organism), it may be possible to apply them to metatranscriptomic data (i.e., obtained from a whole microbial community). 
Given the huge amount of data obtained from metagenomic and metatranscriptomic analysis, the use of bioinformatic tools have become of greater importance in the last decades. In order to achieve so, many different bioinformatic pipelines have been developed, often as open source platforms, such as HUMAnN and the most recent HUMAnN2, MetaTrans, SAMSA and Leimena-2013 
HUMAnN2 is a bioinformatic pipeline designed from the latter HUMAnN developed during the Human Microbiome Project (HMP), implementing a “tiered search” approach. In the first tier, HUMAnN2 screens DNA or RNA reads with MetaPhlAn2 in order to identify already known microbes and constructing a sample-specific database by merging pangenomes of annotated species; in the second tier the algorithm performs a mapping of the reads against the assembled pangenome database; in the third tier, non-aligned reads are used for a translated search against a protein database 
MetaTrans is a pipeline that exploits multi-threading computers to improve metagenomic and metatranscriptomic analysis. Data is obtained from paired-end RNA-Seq, mainly from 16S RNA for taxonomy and mRNA for gene expression levels. The pipeline is divided in 4 major steps. Firstly, paired-end reads are filtered for quality control purposes, to be thereafter sorted for taxonomic analysis (by removal of tRNA sequences) or functional analysis (by removal of both tRNA and rRNA sequencing). For the taxonomic analysis, sequences are mapped against 16S rRNA Greengenes v13.5 database using SOAP2, while for functional analysis sequences are mapped against a functional database such as MetaHIT-2014 always by using SOAP2 tool. This pipeline is highly flexible, since it offers the possibility to use third-party tools and improve single modules as long as the general structure is preserved.
This pipeline is designed specifically for metatranscriptomics data analysis, by working in conjunction with the MG-RAST server for metagenomics. This pipeline is simple to use, requires low technical preparation and computational power and can be applied to a wide range of microbes. The algorithm is divided in 4 steps. At first, sequences from raw sequencing data are selected on quality basis and are then submitted to MG-RAST (which foresee different steps such as quality control check, gene calling, clustering of amino acid sequences and use of sBLAT on each cluster to detect the best matches). Matches are then aggregated for taxonomic and functional analysis purposes, that usually follow up as last steps of the process.
This pipeline actually does not have a name so that it is usually reckoned with the first name of the author of the article in which it is described. This algorithm foresees the implementation of alignment tools such as BLAST and MegaBLAST. Reads, usually obtained by Illumina sequencing, are clustered in identical-reads clusters and are then processed for in-silico removal of t-RNA and r-RNA sequences. Remaining reads are then mapped on NCBI databases by using BLAST and MegaBLAST tools and classified by their bitscore. Higher bitscore sequences are thereby interpreted to predict phylogenetic origin and function. Lower score reads instead are aligned with BLASTX (higher sensitivity) and eventually can be aligned in protein databases so that their function can be characterized.
Another method that can be exploited for metatranscriptomic purposes is Tiling Microarrays. In particular, microarrays have been used to measure microbial transcription levels, to detect new transcripts and to obtain information about the structure of mRNAs (for instance, the UTR boundaries). Recently, it has also been used to find new regulatory ncRNA. However, microarrays are affected by some pitfalls:
- requirement of probe design
- low sensitivity
- prior knowledge of gene targets.
RNA-Seq can overcome these limitations: it does not require any previous knowledge about the genomes that have to be analysed and it provides high throughput validation of genes prediction, structure, expression. Thus, by combining the two approaches it is possible to have a more complete representation of bacterial transcriptome.
Limits of the metatranscriptomic techniques
- With its dominating abundance, ribosomal RNA strongly reduces the coverage of mRNA (main focus of transcriptomic studies) in the total collected RNA.
- Extraction of high-quality RNA from some biological or environmental samples (such as feces) can be difficult
- Instability of mRNA that compromises sample integrity even before sequencing.
- Experimental issues can affect the quantification of differences in expression among multiple samples: They can influence integrity and input RNA, as well as the amount of rRNA remaining in the samples, size section and gene models. Moreover, molecular base techniques are very prone to artefacts.
- Difficulties in differentiating between host and microbial RNA, although com¬mercial kits for the microbial enrichment are available. This may also be done in silico if a reference genome is available for the host.
- Transcriptome reference databases are limited in their coverage.
- Generally, large populations of cells are exploited in metatranscriptomic analysis, so it is difficult to resolve important variances that can exist between subpopulations. Actually, high variability in pathogen populations was demonstrated to affect disease progression and virulence.
- Both for microarray and RNA-Seq, it is difficult to set a real “cut off” in order to consider the genes as “express”, due to the high dynamic range in gene expression.
- The presence of mRNA is not always associated with the actual presence of the respective protein.
Metatrascriptomics and Gut Microbiome
The gut microbiome has emerged in recent years as an important player in human health. Its prevalent functions are related to the fermentation of indigestible food components, competitions with pathogen, strengthening of the intestinal barrier, stimulation and regulation of the immune system. Although much has been learnt about the microbiome community in the last years, the wide diversity of microorganisms and molecules in the gut requires new tools to enable new discoveries. By focusing on the changes in the expression of the genes, metatrascriptomics allows to take a more dynamic picture of the state and activity of the microbiome than metagenomics. It has been observed that metatranscriptomic functional profiles are more variable than what might have been reckoned only by metagenomic information. This suggests that non-housekeeping genes are not stably expressed in situ One example of metatranscriptomic application is in the study of the gut microbiome in inflammatory bowel disease. Inflammatory bowel disease (IBD) is a group of chronic diseases of the digestive tract that affects millions of people worldwide. Several human genetic mutations have been linked to an increased susceptibility to IBD, but additional factors are needed for the full development of the disease. Regarding the relationship between IBD and gut microbiome, it is known that there is a dysbiosis in patients with IBD but microbial taxonomic profiles can be highly different among patients, making it difficult to implicate specific microbial species or strains in disease onset and progression. In addition, the gut microbiome composition presents a high variability over time among people, with more pronounced variations in patient with IBD. The functional potential of an organism, meaning the genes and pathways encoded in its genome, provides only indirect information about the level or extent of activation of such functions. So, the measurement of functional activity (gene expression) is critical to understand the mechanism of the gut microbiome dysbiosis. Alterations in transcriptional activity in IBD, established on the rRNA expression, indicate that some bacterial populations are active in patients with IBD, while other groups are inactive or latent. A metatranscriptomics analysis measuring the functional activity of the gut microbiome reveals insights only partially observable in metagenomic functional potential, including disease-linked observations for IBD. It has been reported that many IBD-specific signals are either more pronounced or only detectable on the RNA level. These altered expression profiles are potentially the result of changes in the gut environment in patients with IBD, which include increased levels of inflammation, higher concentrations of oxygen and a diminished mucous layer. Metatranscriptomics has the advantage of allowing to skip the assaying of biochemical products in situ (like mucus or oxygen) and allows to study the effects of environmental changes on microbial expression patterns in vivo for large human populations. In addition, it can be coupled with longitudinal sampling to associate modulation of activity with the disease progression. Indeed, it has been shown that while a particular path may remain stable over time at the genomic level, the corresponding expression varies with the disease severity. This suggests that microbial dysbiosis affect the gut health through changing in the transcriptional programmes in a stable community. In this way, metatracriptomic profiling emerges as an important tool for understanding the mechanisms of that relationship. Some technical limitations of the RNA measurements in stool are related to the fact that the extracted RNA can be degraded and, if not, it still represents only the organisms presents in the stool sample. Other applications of metagenomics:
- Directed culturing: it was used to understand nutritional preferences of organisms in order to allow the preparation of a proper culture medium, resulting in a successful isolation of microbes in vitro.
- Identify potential virulence factors: through comparative transcriptomics, in order to compare different transcriptional responses of related strains or species after specific stimuli.
- Identify host-specific biological processes and interactions For this purpose, it’s important to develop new technologies which allow the detection, at the same time, of changes in the expression levels of some genes.
Examples of techniques applied: Microarrays: allow the monitoring of changes in the expression levels of many genes in parallel for both host and pathogen. First microarray approaches have shown the first global analysis of gene expression changes in pathogens such as Vibrio cholerae, Borrelia burgdorferi, Chlamydia trachomatis, Chlamydia pneumoniae and Salmonella enterica, revealing the strategies that are used by these microorganisms to adapt to the host. In addition, microarrays only provide the first global insights about the host innate immune response to PAMPs, as the effects of bacterial infection on the expression of various host factor. Anyway, the detection through microarrays of both organisms at the same time could be problematic. Problems:
- Probe selection (hundreds of millions of different probes)
- Need of expensive chips (with the proper design; high-density arrays)
- Require the pathogen and host cells to be physically separated before gene expression analysis (eukaryotic cells’ transcriptomes are larger in comparison to the pathogens’ ones, so could happen that the signal from pathogens’ RNAs is hidden).
- Loss of RNA molecules during the eukaryotic cells lysis.
Dual RNA-Seq: this technique allows the simultaneous study of both host and pathogen transcriptomes as well. It is possible to monitor the expression of genes at different time points of the infection process; in this way could it be possible to study the changes in cellular networks in both organisms starting from the initial contact until the manipulation of the host (interplay host-patogen).
- Potential: No need of expensive chips
- Probe-independent approach (RNA-seq provides transcript information without prior knowledge of mRNA sequences)
- High sensitivity.
- Possibility of studying the expression levels of even unknown genes under different conditions
Moreover, RNA-Seq is an important approach for identifying coregulated genes, enabling the organization of pathogen genomes into operons. Indeed, genome annotation has been done for some eukaryotic pathogens, such as Candida albicans, Trypanosoma brucei and Plasmodium falciparum. Despite the increasing sensitivity and depth of sequencing now available, there are still few published RNA-Seq studies concerning the response of the mammalian host cell to the infection.
- Filiatrault MJ (October 2011). "Progress in prokaryotic transcriptomics". Current Opinion in Microbiology. 14 (5): 579–86. doi:10.1016/j.mib.2011.07.023. PMID 21839669.
- Bashiardes S, Zilberman-Schapira G, Elinav E (2016). "Use of Metatranscriptomics in Microbiome Research". Bioinformatics and Biology Insights. 10: 19–25. doi:10.4137/BBI.S34610. PMC 4839964. PMID 27127406.
- Whipps JM, Lewis K, Cooke RC (1988). Mycoparasitism and Plant Disease Control. Manchester, UK: Manchester University Press. pp. 161–87.
- Moran MA (2009). "Metatranscriptomics: eavesdropping on complex microbial communities". Microbiome. 4 (7): 329–34.
- Apirion D, Miczak A (February 1993). "RNA processing in prokaryotic cells". BioEssays : News and Reviews in Molecular, Cellular and Developmental Biology. 15 (2): 113–20. doi:10.1002/bies.950150207. PMID 7682412.
- Peimbert M, Alcaraz LD (2016). "A Hitchhiker's Guide to Metatranscriptomics". Field Guidelines for Genetic Experimental Designs in High-Throughput Sequencing. Springer. pp. 313–342.
- Dumont MG, Pommerenke B, Casper P (October 2013). "Using stable isotope probing to obtain a targeted metatranscriptome of aerobic methanotrophs in lake sediment". Environmental Microbiology Reports. 5 (5): 757–64. doi:10.1111/1758-2229.12078. PMID 24115627.
- Grabherr MG, Haas BJ, Yassour M, Levin JZ, Thompson DA, Amit I, et al. (May 2011). "Full-length transcriptome assembly from RNA-Seq data without a reference genome". Nature Biotechnology. 29 (7): 644–52. doi:10.1038/nbt.1883. PMC 3571712. PMID 21572440.
- 103. Li B, Dewey CN. Rsem: accurate transcript quantification from RNA-seq data with or without a reference genome. BMC Bioinformatics. 2011;12(1):323.
- Haas BJ, Papanicolaou A, Yassour M, Grabherr M, Blood PD, Bowden J, Couger MB, Eccles D, Li B, Lieber M, MacManes MD, Ott M, Orvis J, Pochet N, Strozzi F, Weeks N, Westerman R, William T, Dewey CN, Henschel R, LeDuc RD, Friedman N, Regev A (August 2013). "De novo transcript sequence reconstruction from RNA-seq using the Trinity platform for reference generation and analysis". Nature Protocols. 8 (8): 1494–512. doi:10.1038/nprot.2013.084. PMC 3875132. PMID 23845962.
- De Bona F, Ossowski S, Schneeberger K, Rätsch G (August 2008). "Optimal spliced alignments of short sequence reads". Bioinformatics. 24 (16): i174–80. doi:10.1093/bioinformatics/btn300. PMID 18689821.
- Leimena MM, Ramiro-Garcia J, Davids M, van den Bogert B, Smidt H, Smid EJ, Boekhorst J, Zoetendal EG, Schaap PJ, Kleerebezem M (2013). "A comprehensive metatran¬scriptome analysis pipeline and its validation using human small intestine micro¬biota datasets". BMC Genomics. 14 (1): 530.
- Yost S, Duran-Pinedo AE, Teles R, Krishnan K, Frias-Lopez J (December 2015). "Functional signatures of oral dysbiosis during periodontitis progression revealed by microbial metatranscriptome analysis". Genome Medicine. 7 (1): 27. doi:10.1186/s13073-015-0153-3. PMC 4410737. PMID 25918553.
- Yost S, Duran-Pinedo AE, Teles R, Krishnan K, Frias-Lopez J (2015). "Functional signatures of oral dysbiosis during periodontitis progression revealed by microbial metatranscriptome analysis". Genome Medicine. 7 (1): 27. doi:10.1186/s13073-015-0153-3. PMC 4410737. PMID 25918553.
- Duran-Pinedo AE, Chen T, Teles R, Starr JR, Wang X, Krishnan K, Frias-Lopez J (August 2014). "Community-wide transcriptome of the oral microbiome in subjects with and without periodontitis". The ISME Journal. 8 (8): 1659–72. doi:10.1038/ismej.2014.23. PMC 4817619. PMID 24599074.
- Jorth P, Turner KH, Gumus P, Nizam N, Buduneli N, Whiteley M (April 2014). "Metatranscriptomics of the human oral microbiome during health and disease". mBio. 5 (2): e01012–14. doi:10.1128/mBio.01012-14. PMC 3977359. PMID 24692635.
- Xiong X, Frank DN, Robertson CE, Hung SS, Markle J, Canty AJ, McCoy KD, Macpherson AJ, Poussier P, Danska JS, Parkinson J (2012). "Generation and analysis of a mouse intestinal metatranscriptome through Illumina based RNA-sequencing". PLOS ONE. 7 (4): e36009. doi:10.1371/journal.pone.0036009. PMC 3338770. PMID 22558305.
- Niu SY, Yang J, McDermaid A, Zhao J, Kang Y, Ma Q (November 2018). "Bioinformatics tools for quantitative and functional metagenome and metatranscriptome data analysis in microbes". Briefings in Bioinformatics. 19 (6): 1415–1429. doi:10.1093/bib/bbx051. PMID 28481971.
- Franzosa EA, McIver LJ, Rahnavard G, Thompson LR, Schirmer M, Weingart G, Lipson KS, Knight R, Caporaso JG, Segata N, Huttenhower C (November 2018). "Species-level functional profiling of metagenomes and metatranscriptomes". Nature Methods. 15 (11): 962–968. doi:10.1038/s41592-018-0176-y. PMC 6235447. PMID 30377376.
- Martinez X, Pozuelo M, Pascal V, Campos D, Gut I, Gut M, Azpiroz F, Guarner F, Manichanh C (May 2016). "MetaTrans: an open-source pipeline for metatranscriptomics". Scientific Reports. 6: 26447. doi:10.1038/srep26447. PMC 4876386. PMID 27211518.
- Westreich ST, Korf I, Mills DA, Lemay DG (September 2016). "SAMSA: a comprehensive metatranscriptome analysis pipeline". BMC Bioinformatics. 17 (1): 399. doi:10.1186/s12859-016-1270-8. PMC 5041328. PMID 27687690.
- Leimena MM, Ramiro-Garcia J, Davids M, van den Bogert B, Smidt H, Smid EJ, Boekhorst J, Zoetendal EG, Schaap PJ, Kleerebezem M (August 2013). "A comprehensive metatranscriptome analysis pipeline and its validation using human small intestine microbiota datasets". BMC Genomics. 14: 530. doi:10.1186/1471-2164-14-530. PMC 3750648. PMID 23915218.
- Karasov WH, Martínez del Rio C, Caviedes-Vidal E (2011). "Ecological physiology of diet and digestive systems". Annual Review of Physiology. 73: 69–93. doi:10.1146/annurev-physiol-012110-142152. PMID 21314432.
- LeBlanc JG, Milani C, de Giori GS, Sesma F, van Sinderen D, Ventura M (April 2013). "Bacteria as vitamin suppliers to their host: a gut microbiota perspective". Current Opinion in Biotechnology. 24 (2): 160–8. doi:10.1016/j.copbio.2012.08.005. PMID 22940212.
- Claus SP, Guillou H, Ellero-Simatos S (2016). "The gut microbiota: a major player in the toxicity of environmental pollutants?". NPJ Biofilms and Microbiomes. 2: 16003. doi:10.1038/npjbiofilms.2016.3. PMC 5515271. PMID 28721242.
- Kamada N, Seo SU, Chen GY, Núñez G (May 2013). "Role of the gut microbiota in immunity and inflammatory disease". Nature Reviews. Immunology. 13 (5): 321–35. doi:10.1038/nri3430. PMID 23618829.
- Abreu MT (February 2010). "Toll-like receptor signalling in the intestinal epithelium: how bacterial recognition shapes intestinal function". Nature Reviews. Immunology. 10 (2): 131–44. doi:10.1038/nri2707. PMID 20098461.
- Sommer F, Bäckhed F (April 2013). "The gut microbiota--masters of host development and physiology". Nature Reviews. Microbiology. 11 (4): 227–38. doi:10.1038/nrmicro2974. PMID 23435359.
- Hooper LV, Littman DR, Macpherson AJ (June 2012). "Interactions between the microbiota and the immune system". Science. 336 (6086): 1268–73. doi:10.1126/science.1223490. PMC 4420145. PMID 22674334.
- Gosalbes MJ, Durbán A, Pignatelli M, Abellan JJ, Jiménez-Hernández N, Pérez-Cobas AE, Latorre A, Moya A (March 2011). "Metatranscriptomic approach to analyze the functional human gut microbiota". PLOS ONE. 6 (3): e17447. doi:10.1371/journal.pone.0017447. PMC 3050895. PMID 21408168.
- Franzosa EA, Morgan XC, Segata N, Waldron L, Reyes J, Earl AM, Giannoukos G, Boylan MR, Ciulla D, Gevers D, Izard J, Garrett WS, Chan AT, Huttenhower C (June 2014). "Relating the metatranscriptome and metagenome of the human gut". Proceedings of the National Academy of Sciences of the United States of America. 111 (22): E2329–38. doi:10.1073/pnas.1319284111. PMC 4050606. PMID 24843156.
- Burisch J, Jess T, Martinato M, Lakatos PL (May 2013). "The burden of inflammatory bowel disease in Europe". Journal of Crohn's & Colitis. 7 (4): 322–37. doi:10.1016/j.crohns.2013.01.010. PMID 23395397.
- Halfvarson J, Brislawn CJ, Lamendella R, Vázquez-Baeza Y, Walters WA, Bramer LM, D'Amato M, Bonfiglio F, McDonald D, Gonzalez A, McClure EE, Dunklebarger MF, Knight R, Jansson JK (February 2017). "Dynamics of the human gut microbiome in inflammatory bowel disease". Nature Microbiology. 2 (5): 17004. doi:10.1038/nmicrobiol.2017.4. PMC 5319707. PMID 28191884.
- Lewis JD, Chen EZ, Baldassano RN, Otley AR, Griffiths AM, Lee D, et al. (October 2015). "Inflammation, Antibiotics, and Diet as Environmental Stressors of the Gut Microbiome in Pediatric Crohn's Disease". Cell Host & Microbe. 18 (4): 489–500. doi:10.1016/j.chom.2015.09.008. PMC 4633303. PMID 26468751.
- Rehman A, Lepage P, Nolte A, Hellmig S, Schreiber S, Ott SJ (September 2010). "Transcriptional activity of the dominant gut mucosal microbiota in chronic inflammatory bowel disease patients". Journal of Medical Microbiology. 59 (Pt 9): 1114–22. doi:10.1099/jmm.0.021170-0. PMID 20522625.
- Naughton J, Duggan G, Bourke B, Clyne M (2014). "Interaction of microbes with mucus and mucins: recent developments". Gut Microbes. 5 (1): 48–52. doi:10.4161/gmic.26680. PMC 4049936. PMID 24149677.
- Westermann AJ, Gorski SA, Vogel J (September 2012). "Dual RNA-seq of pathogen and host". Nature Reviews Microbiology. 10 (9): 618–630. doi:10.1038/nrmicro2852. PMID 22890146.
- Saliba AE, C Santos S, Vogel J (February 2017). "New RNA-seq approaches for the study of bacterial pathogens". Current Opinion in Microbiology. 35: 78–87. doi:10.1016/j.mib.2017.01.001. hdl:10033/621506. PMID 28214646.