METAGENOMICS AND METATRANSCRIPTOMICS OF A H2S
DESULFURIZING BIOTRICKLING FILTER
TERCIA BEZERRA*, ROGER ROVIRA*, ANDREA MONTEBELLO*
MONTSERRAT LLAGOSTERA**, JAVIER LAFUENTE*, SUSANA CAMPOY**,
*Department of Chemical Engineering, Universitat Autònoma de Barcelona, Barcelona, Spain
**Department of Genetics and Microbiology, Universitat Autònoma de Barcelona, Barcelona, Spain
Corresponding author: email@example.com, +34935811587
The application of "bio-omics" has been crucial to the development of environmental microbiology
mostly for its comprehensive coverage. In this work, the 454 Roche platform was applied for massive
sequencing of nucleic acids for studying metagenomics and metatranscriptomics of a biotrickling filter
for H2S desulfurization under aerobic conditions. The dynamics of microbial community diversity was
studied before and after a pH change from 6.5 to 2.5. The metatranscriptome was only studied under
neutral pH conditions. The analysis of relative abundance data revealed that desulfurization was
carried out by a quite diverse bacterial community under neutral pH. However, after the pH change an
extinction of many species occurred while prevalence of two species in the community was observed;
Acidithiobacillus thiooxidans and Acidiphilim sp. Results indicated that massive sequencing of 16S
rRNA may help understanding the effects of changes in operational conditions on ecological complex
systems. The metatranscriptomic approach permitted to find around 3000 genes that were being
expressed in the desulfurizing biotrickling filter under neutral pH operation. Expression of such genes
along the biotrickling filter height was quantified. However, data analysis was limited due to limited
information available in current databases. Despite such limitations, some genes were related with
proteins involved in the sulfur metabolism.
Keywords: Microbial diversity, gene expression, massive sequencing, biotrickling filter, biogas
Classical molecular biology techniques have helped revealing the composition and distribution of the
microbiota that develops in complex environments (Maestre et al., 2010). However, these studies have
failed to establish the interactions and functions of species within the community. For this reason, the
structure-function paradigm of complex microbial communities remains one of the biggest challenges
of microbial ecology and environmental bioprocesses. Recently, new approaches have emerged
applicable to the study of the microbial communities at different levels. This is due to advances in new
sequencing platforms and computational analyzes aimed at interpreting the data. Such approaches, so-
called "bio-omics", usually share three major characteristics that contrast with the classical molecular
biology techniques. Bio-omics are high performance methodologies based on the massive data analysis
and with a holistic approach of the system. From the new point of view, complex biological systems
are integral systems instead of merely partial collections based on information related sequence (Ishi
and Tomita, 2009). Under this concept, various research areas including genomics, transcriptomics,
proteomics, metabolomics and interactomics have appeared. These lines of research are considering
each of the cellular levels from a global perspective and integral biological system (Hertog et al.,
Through metagenomics, the microbial diversity has been studied in a quantitative manner based on
massive sequencing of all genomes of a given environment (Ghai et al., 2011). Similarly, the
metatranscriptomic analysis has enabled contemplate quantitatively and integrally the gene expression
for complex communities. This type of analysis shows the expression of all genes active at a time and
particular environmental condition (Zhanget al., 2010). Although scarce, some studies have been
reported about the metatranscriptome and the metagenome of bioreactors for the production of biogas
(Zakrzewski et al., 2012) and biofilters for H2S removal (Li et al., 2011).
In previous works, we used molecular biology techniques such as cloning and sequencing and
Fluorescence In-Situ Hybridization (FISH) for the characterization of populations in a desulfurizing
biotrickling filter (Maestre et al., 2010) and their evolution over time in different regions of the reactor
and under different operating conditions. However, results indicated that these techniques are slow,
expensive, and not very repetitive and partially cover the microbial diversity colonizing the reactor.
Based on these limitations we decided to carry out a study to deepen the knowledge of the biotic part
of the H2S biofiltration process. New platforms for massive DNA sequencing have been used.
Specifically, pyrosequencing in the Roche 454 platform has been used in this work to assess the
metagenome (microbial diversity) and metatranscriptome (gene expression) of a desulfurizing
biotrickling filter. Analyses have been used to develop a reliable, repetitive technique having a deep
coverage to identify microbial populations. Also, sequencing was used to determine the gene
expression of bacterial communities under certain micro environmental conditions. Metagenome was
assessed in samples obtained at pH 7 and 2.5 while the metatranscriptome was assessed under neutral
pH operation of the bioreactor.
MATERIALS AND METHODS
Operational conditions of the biotrickling biofilter
The experimental setup consisted of a conventional counter-current biotrickling filter, operated in up
flow mode, fully monitored and automated. A biogas mimics flow is fed at the lower section of the
reactor, packed with 2 L of a conventional random packing material (Pall rings, AISI 316, 10 mm
diameter). Oxygen was supplied by bubbling air into an auxiliary unit in the liquid recirculation line.
Except for some short experiments, the trickling biofilter was operated continuously for more than 900
days at a constant inlet H2S concentration of 2000 ppmv (load of 55.6 g H2S m-3h-1). Removal
efficiency was usually greater than 95% during the whole experimental period independently of the
operational pH. A detailed description of both the bioreactor configuration and operational conditions
can be found elsewhere (Montebello et al., 2012; Rovira, 2012). The pH was gradually reduced from
6.5 to 2.5 after 440 days of operation. Biomass was collected at three different bed heights located at
40 cm (upper region), 23 cm (middle) and 4 cm (bottom) of a total length of 53 cm. After startup,
samples were taken on day 350 (metatranscriptome sample), and in days 245 and 586 for metagenome
Biomass samples collected in the 3 sections along the length of the reactor were mixed. Samples were
taken on day 245 (neutral pH, steady-state conditions) and on day 586 (acidic pH, steady-state
conditions). Biomass attached to the packing material and to the reactor internal wall was displaced by
mechanical agitation in 30 mM PBS buffer and then concentrated by centrifugation at 7500 rpm for 15
minutes. The supernatant was discarded and the pellet was preserved at -20 ºC for further DNA
extraction. 0.25 g of pellets was used for DNA extraction by Power soil DNA isolation kit (Mo Bio
Laboratories Inc., Carlsbad, CA, USA) according to the manufacturer instructions. The quality and
concentration of extracted DNA were checked in a Nanodrop (Thermo Fisher Scientific, MA, USA) in
2 µl of each DNA elution. The ratio 1.8 at 260/280 nm was used as quality cut. Concentrations of
samples were normalized at 20 ng/µl. Also, DNA was amplified for pyrosequencing using forward and
reverse fusion primers. Amplicon libraries were constructed using 16S universal Eubacteria primers.
The forward primer was constructed with (5’-3’) Roche A linker, an 8-10 bp barcode and the 341
primer (5’CCT ACG GGA GGC AGC AG 3’). The reverse fusion primer was constructed with (5’-3’)
a biotin molecule, the Roche B linker, and the 907 primer (5’CCG TCA ATT CMT TTG AGT TT3’).
Amplifications were performed in 25 µl reactor volumes with Qiagen HotStar Taq master mix (Qiagen
Inc., CA, USA), 1 µl of each 5 µM primer and 1 µl of template. Reactions were performed on ABI
Veriti Thermocyclers (Applied Biosytems, Carlsbad, CA, USA) under the following thermal profile:
95 ○C for 5 min., then 35 cycles of 94 ○C for 30 s, 54 ○C for 40 s, 72 ○C for 1 min followed by one
cycle of 72 ○C for 10 min.. Amplification products were visualized with Gels (Life Technologies,
Grand Island, New York). Products were then pooled equimolar and each pool was cleaned with
Diffinity RapidTip (Diffinity Genomics, West Henrietta, NY, USA). Size was selected using
Agencourt AMPure XP (Beckman Coulter, Indianapolis, Indiana, USA) following the Roche 454
protocols (454 Life Sciences, Branford, Connecticut, USA). Size selected pools were then quantified
and 150 ng of DNA were hybridized to Dynabeads M-270 (Life Technologies) to create single
stranded DNA following Roche 454 protocols (454 Life Sciences). Single stranded DNA was diluted
and used in PCR reactions, which were performed and subsequently enriched.
Sequencing was performed on a FLX+ 454 Roche platform at Research and Testing Laboratory (RTL,
Texas, USA) following established manufacturer protocols (454 Life Sciences). Sequences
identification was also performed at RTL. All methods are detailed in RTL website (RTL, 2012). The
quality of sequencing data obtained was evaluated and denoised. Chimeras were removed from the
data set. Briefly, USEARCH was used for denoise checking (Edgar, 2010). Data were de-replicated
and clustered (at 1% divergence). Also, reads shorter than 267 bp and reads larger than 603 bp were
trimmed. After this process, consensus sequences were generated and re-sorted. Then, all poor reads
were eliminated. The detection and removal of chimeras was made on UCHIIME in de novo mode on
the clustered data. Later, a file was generated in FASTA format and original tags were replacement by
artificial tags for taxonomic identification of the reads. To determine the identity, the remaining
sequences were clustered in OTUS with 96.5% using USEARCH. The FASTA file with seeded
sequences was then queried against the NCBI database utilizing BLASTN+ according to previously
established methods (Andreoti et al., 2011). Based on the sequence identity percentage criteria,
sequences were classified at appropriate taxonomic levels. The following percentage criteria were
used: greater than 97% identity at species level; between 95% and 97% at genus level; between 90%
and 95% at family level; between 85% and 90% at order level; 80 and 85% at class level; and 77% to
80% at phyla level. Later, the abundance percentage was individually calculated for each sample. This
provided relative abundance data within and among the individual samples based upon the relative
numbers of reads during sequencing.
Metatranscriptomic analysis was performed after mRNA extraction of samples obtained from each
region (upper, middle and bottom) after 350 days of operation of the biotrickling filter. Complete
protocols can be found in Rovira (2012). In short, each sample was dissolved in PBS buffer and
pelleted. The supernatant was discarded while pellets were processed immediately. Total RNA from
the samples was isolated with PowerSoil Total RNA Isolation Kit (Mobio) and total RNA from each
sample was incubated with DNase (TURBO DNA-free ™ Kit, Ambion/Applied Biosystems,
Darmstadt, Germany) to remove traces of DNA. Later, two enrichment cycles with MICROBExpress
™ Kit (Ambion) were performed and the resulting mRNA was purified with the MEGAclear ™ Kit
(Ambion). The preparation method of the cDNA library was adapted from the Roche manual "cDNA
Rapid Library Preparation Method Manual "for the GS FLX Titanium Series platform (October 2009).
A maximum volume of 19 µl of sample was used with a minimum purity of A260/280 ≥ 1.8 and a
minimum of 200 ng of mRNA. Double-stranded cDNA synthesis (dscDNA) was performed from
mRNA obtained of the 3 samples. In short, 4 ml of random primers were added (400 mM Prime
Random Roche) and resulting volumes were incubated at 70 ºC for 10 min. Then, samples were
transferred on ice and processed to the synthesis of cDNA using the cDNA System Synthesis Kit
(Roche) according to manufacturer protocol. The dscDNA generated was prepared for pyrosequencing
with the binding of adapters A and B, and tags per each sample. In this work the quality and the
concentration of nucleic acids were examined after each step in a Nanodrop ND-1000
(ThermoScientific) or Agilent 2100 Bioanalyzer (Agilent Technologies).
After pyrosequencing, the optimum assembly strategy for studying the metatranscriptome of the
desulfurizing biotrickling filter was defined from reads generated during sequencing. A de novo
assemblage strategy defined as "Genomic Project de novo assembly" was chosen. Tracing the origin of
each reading served to assess the gene expression in each section of the bioreactor. Moreover,
expression levels of each contig were normalized and quantified by statistical analysis using RPKM
(Reads per Kilobase per Million mapped reads) (Mortazavi et al. 2008). Then, contigs were filtered to
identify the genes that showed a marked differential expression. In this case, contigs with RPKM
exceeding 10% of the average RPKM were eliminated. After, an analysis of phylogenetic and
functional annotations of the contigs with a degree of unequal expression between different
biotrickling filter sections was performed. Search for similarity and associated functionality of species
was performed with the BLASTX program, comparing the six possible reading frames of the
sequences against nucleotide database of non-redundant protein sequences (NR) in NCBI
(http://blast.ncbi.nlm.nih.gov/). We defined a maximum e-value of 10-3 as a condition to proceed with
the functional characterization. The best entry was selected for each contig (lowest e-value). In
parallel, a search by similarity was conducted specifically for the small subunit rRNA
(SSURef_106_NR) with BLASTn against SILVA database (www. arb-silva.de/). An e-value of 10-5
was defined with a minimum length of 65 nt and a minimum similarity index of 80%. Functional
characterization of the sequences was performed in Blast2Go (Conesa et al., 2005). Furthermore, reads
of sequences were contrasted with the KEEG database (Kanehisa et al., 2011) to provide a metabolic
map of the annotated sequences.
RESULTS AND DISCUSSION
Microbial community composition under neutral and acidic pH conditions
The metagenomic analysis of Eubacteria community in the H2S desulfurizing biotrickling biofilter was
performed in two samples taken before (day 245 after startup) and after (day 586 after startup) a
change of pH from neutral to acidic. Two libraries, namely 245d and 586d, were constructed. A total
of 67994 sequence reads were obtained by sequencing. After quality analysis and removal of low
quality sequences, 49822 sequences were annotated (73% of initial); 12869 reads in library 245d and
36953 reads in library 586d. Taxonomic identification of each read sequence was investigated and the
relative abundance data (number of sequences by sample) were calculated for each library. Overall,
4381 OTUs were assigned. A total of 2233 and 1789 OTUs were obtained in libraries 245d and 586d,
respectively. All data was grouped in 265 entries (OTUs grouped by similarity with reference
sequences). In this study, species similarity was found for a total of 129 clusters.
A deep change in the diversity was found between both sampling events analyzed. The change in the
community structure is presented in Fig. 1. Briefly, 73 families were found which were distributed in
19 classes. All 19 classes were present in library 245d while only 9 classes were found in library 586d.
Relative abundance of reads in library 245d indicated that the most abundant families were
Xanthomonadaceae (γ) with 35% and Nitrosomonadaceae (β) with 22.5%. Also, relative abundances
showed that 73 families fail to reach to 1% abundance. Regarding library 586d, complete
disappearance of Betaproteobacteria (β) was found, while Gammaproteobacteria (γ) (82%) and
Alphaproteobacteria (α) (12%) prevailed after the pH change (see Fig 1).
Figure 1 Bacterial community composition of the samples taken on day 245(neutral pH) and on day
586 (acid pH) at class taxonomic level. The relative abundance of prevalent species in the respective
class is shown.
Changes in the community structure of Alphaproteobacteria were particularly interesting to illustrate
the effect of pH on the community. Briefly, in library 245d 58 OTUs (α) showed very reduced relative
abundances (up to 2%), which indicates a balanced distribution of the populations during the period
operated at neutral pH. Contrary to this, only 13 OTUs (α) were noted in library 586d. Most important,
95% of sequences were similar to Acidiphilium sp. This dramatic change clearly occurred in response
to environmental changes. Data indicates specialization of the community and the prevalence of
species adapted to very low pH. It is also important to note that the abundance at class level was not
very different in the two libraries (around 12%) but the diversity changed markedly, which means that
class level data were not sufficient to allow investigating dynamics of the community. With respect to
the Betaproteobacteria extinction (abundance felt from 36% to less than 0.5%), this can be partly
related to the disappearance of Nitrosomonadaceae (relative abundance of 22% in 245d). This situation
similarly occurred to sulfide-oxidizing bacteria (SOB) populations. Neutrophilic species of the genus
Thiobacillus and Thiomonas felt to a 6% of relative abundance. The success of Gammaproteobacteria
at acid pH was attributed to the species of the genus Acidithiobacillus, which increased from 5% to
60% (prevalent group) between neutral and acidic pH. Family Xanthomonadaceae kept at around 30%.
Moreover, such maintenance was due to the growth of Dyella ginsengisoli that increased from 3% to
17%. Also, 35 of the 43 species were eliminated or reduced to less than 1% under acid pH. This data
illustrate the selective pressure of the acidification of microenvironment.
Global analysis of metatranscriptome under neutral pH conditions
Initially, we proceeded with sorting and filtering the generated contigs. The assembly was made from
the set of readings after running a N50 statistical analysis. Results of this analysis indicated that the
best assembly (more efficient and higher quality) was made in the full data set (data not showed). The
3181 resulting contigs from the reconstructions were categorized according to the origin of the sample.
Approximately 82.1% of the generated contigs appeared in the three sections of the biofilter, 15.4%
were identified only in two sections of the bioreactor and finally only 2.6% were specific to a single
section of the bioreactor. The differential expression analysis and comparison of global expression
levels (RPKM and Spearman indexes) showed that, comparatively, samples from the upper and half
regions of the reactor shared similar overall expression levels (Spearman coefficient of 0.92). In
comparison, the upper and lower zones of the biotrickling filter showed a Spearman coefficient of
0.87, indicating a greater degree of differentiation between them. The Spearman coefficient for middle
and lower regions was low (0.84) which shows that the transcriptomes were more divergent. In the
assembly of contigs obtained, 2749 (about 86.5%) of the total showed at least one entry in the search
for similarity in the protein sequences NR (NCBI) database. Out of them, only 2681 showed scores
according to the defined cutoff criteria (e-value ≤·10-3). For each of these contigs, the best input was
selected and annotated from which analysis and functional characterization was performed.
Nucleotide-nucleotide comparative search was made in rRNA (SILVA). Only 993 contigs showed
sequence similarity in this database, exceeding the cutoff criteria defined. Also, 2681 contigs were
submitted to Blast2GO of which yielded the access number of the respective protein. With this, an
ontological characterization and description of proteins was performed. At the end, ontological
descriptions were analyzed in 102 contigs of the global metatranscriptome. Figure 2B shows the
distributions of the terms listed in ontological categories GO, molecular function. Still, some
functional annotation analysis obtained was probably not reliable to reality functional desulfurization
of the biomass of bioreactor.
Figure 2 Heat map of the three zones (low, medium and up) (A). The gradation of grey represents the
level of expression of all the three samples considered contigs, 0 (white) being the lowest and 7 (dark
grey) the expression level of maximum considered. The lateral dendrogram represents sequence
similarity between contigs considered, while the upper dendrogram shows the distance between the
global expression of three samples considered: low zone (rpkmLow), middle zone (rpkmMedium),
upper (rpkmUp). In “B” distribution of ontological characterizations obtained in the category of
The data presented herein have been linked with the operation of the biofilter. Some contigs found
corresponded to specific proteins of Thiobacillus denitrificans and Sulfurimonas denitrificans. These
species have been previously identified in the same biotrickling filter through a cloning and
sequencing approach (Maestre et al., 2010; Rovira et al, 2010). Furthermore, a variety of other proteins
showed affiliation with SOB such as Acidithiobacillus, Chlorobi and a large number of uncultivable
clones. Also several specific proteins associated with Sulfuricurvum kujiense appeared as well as a
hypothetical protein of Sulfurihydrogenibium azorense. Furthermore, some contigs associated with
Rpkm Low Medium Up
1 Structural molecule activity
2 Transporter activities
3 Molecular transducer activities
4 Electron carrier activities
5 Catalytic activities
specific proteins of species of the domain Archaea such as Sulfolobus solfataricus were identified.
Several proteins of Desulfovibrio and other uncultivated clones of Desulfobacterium, all related with
sulfate-reduction, were also identified. This fact points out the coexistence of sulfate reduction and
oxidation in H2S desulfurization bioreactors, at least in those were anaerobic conditions prevail in
some sections of the reactor or inner parts of the biofilm. Overall assessment of the functional
annotation of the metatranscriptome obtained from the bioreactor highlights the lack of biological and
sequence data available about the microbial communities that develop in H2S desulfurizing
biotrickling filters. Despite the growing number of publications about metagenomic and
metatranscriptomics (Nonnenmann et al. 2010; Li et al. 2011; Lu et al. 2012; Zakrzewski et al. 2012)
little has been clarified about genetic distribution of sulfur metabolism across taxa. However, a great
diversity of metabolic strategies has been identified. Also, the wide distribution of the ability of
metabolizing different forms of sulfur is due to the evolutionary divergence of microorganisms. Even
more, the versatility observed in the small fraction of the diversity that is known is used to imagine the
great challenge that represents research through “bio-omics” approaches.
The metagenomic approach indicated that the relative abundance at family or class levels by
themselves were not enough to clarify the dynamics of the microbial community of a H2S
desulfurizing biotrickling filter. Genera and species annotation is necessary to investigate the
community in the biofilter. The data revealed a large complexity of the bacterial community which
developed in the biofilter during the operational period at neutral pH, while a large diversity reduction
occurred at acidic pH, even if the desulfurizing capacity of the reactor remained unaffected.
The metatranscriptomic approach permitted a quantitative analysis of the overall expression of genes
which suggested that the metabolic activity was similar in all three regions of the biotrickling filter.
Furthermore, comparative analysis of the three regions indicated that the pollutant-degrading activity
takes place in the part closest to the entrance of the gas in the reactor. Expression of genes throughout
the body of the reactor were more similar, probably because the dynamic recirculation of the liquid
phase. Still, in future work it is necessary to identify and quantify the expression of genes in biological
function involved in sulfide oxidation mechanisms since a lack of functional annotation was found in
databases. It would be interesting, for example, to deeply analyze expression profiles of each region of
the reactor, focusing the analysis on those contigs that show a more marked differential expression, as
well as to quantify the levels of expression of genes related to specific metabolic pathways, such as the
Sox cluster. This could help identify more genes linked to the environmental conditions of the
bioreactor. The study of expression should also be carried out under different operating conditions to
increase knowledge of the functional potential of sulfide-oxidizing communities.
The Spanish government (MEC) provided financial support through the project CICYT CTM2009-
14338-C03-01 and CTM2009-14338-C03-02. The Department of Chemical Engineering at UAB
(Universitat Autònoma de Barcelona) is an unit of Biochemical Engineering of the Xarxa de
Referència en Biotecnologia de Catalunya (XRB), Generalitat de Catalunya.
Andreotti R., León A. P. de, Dowd S., Guerrero F. ,Bendele K., Scoles G. (2011). Assessment of
bacterial diversity in the cattle tick Rhipicephalus (Boophilus) microplus through tag-encoded
pyrosequencing, BMC Microbiology, 11, 6.
Conesa A., Götz S., García-Gómez J. M., Terol J., Talón M., Robles M., (2005). Blast2GO: a universal
tool for annotation, visualization and analysis in functional genomics research, Bioinformatics 21,
Edgar R. C. (2010). Search and clustering orders of magnitude faster than BLAST, Bioinformatics, 12,
Ghai R., Rodriguez-Valera F., Mcmahon, K. D., Toyama D., Rinke R., Oliveira T. C. S. de, Garcia J.
W., Miranda F. P. de, Henrique-Silva, F. (2011). Metagenomics of the water column in the pristine
upper course of the Amazon river, PLoS ONE, 6, 8. e23785.
Hertog M.L. A. T. M., Rudell D. R., Pedreschi R., Schaffer R. J., Geeraerd A. H., Nicolai B. M.,
Ferguson I. (2011). Where systems biology meets postharvest, Postharvest Biology and
Technology, 62 (3), 223-237.
Ishii N., Tomita M. (2009). Multi-Omics Data-Driven Systems Biology of E. coli. In: Systems Biology
and Biotechnology of Escherichia coli, S. Y. Lee (Eds.). Springer Netherlands, pp 41-57.
Kanehisa M., Goto S., Sato Y., Furumichi M., Tanabe, M. (2011). KEGG for integration and
interpretation of large-scale molecular data sets, Nucleic Acids Research., 40(D1),109-114.
Li J., Ye G., Sun D., Sun G., Zeng X., Xu J., Liang S. (2012). Performances of two biotrickling filters
in treating H2S-containing waste gases and analysis of corresponding bacterial communities by
pyrosequencing. Appl. Microbiol. Biotechnol., 95(6), 1633-41.
Lu L., Xing D., Ren N. (2012). Pyrosequencing reveals highly diverse microbial communities in
microbial electrolysis cells involved in enhanced H2 production from waste activated sludge, Water
Research, 46(7), 2425-2434.
Maestre J. P., Rovira R., Álvarez-Hornos F. J., Fortuny, M., Lafuente, J. X., Gamisans, D. G. (2010).
Bacterial community analysis of a gas-phase biotrickling filter for biogas mimics desulfurization
through the rRNA approach, Chemosphere, 80, 872–880.
Montebello A. M., Fernández M., Almenglo F., Ramírez M., Cantero D., Baeza M., Gabriel D.
(2012). Simultaneous methylmercaptan and hydrogen sulfide removal in the desulfurization of
biogas in aerobic and anoxic biotrickling filters, Chemical Engineering Journal, 200 (15), 237-246.
Mortazavi A., Williams B. A., McCue K., Schaeffer, L., Wold, B., (2008). Mapping and
quantifying mammalian transcriptomes by RNA-Seq, Nat Meth 5, 621-628.
Nonnenmann M. W., Bextine B., Dowd S. E., Gilmore K., Levin J. L. (2010) Culture-independent
characterization of bacteria and fungi in a poultry bioaerosol using pyrosequencing: a new
approach, J. Occup. Environ. Hyg., 7(12), 693-9.
Rovira R., Montebello A. M., Valle A., Gómez J. M., Langostera M., Lafuente J., Campoy S., Gabriel
D. (2010). Advanced tools in microbial characterization and gene expression of a biotrickling filter
treating high loads of H2S. In: Proceedings of the 2010 Duke-UAM Conference on Biofiltration on
CD-ROM. Washington, USA.
Rovira R. (2012). Characterization of the operation and study of the metatranscriptome of a bioreactor
for desulfurization of high loading of H2S. PhD thesis (In Catalan). Universitat Autònoma de
RTL (2012) http://www.researchandtesting.com/docs/Data_Analysis_Methodology.pdf. Last accessed
Zakrzewski M., Goesmann A., Jaenicke S., Jünemann S. , Eikmeyer F., Szczepanowski R., Al-Soud
W.A., Sørensen S., Pühler A., Schlüter A. (2012). Profiling of the metabolically active community
from a production-scale biogas plant by means of high-throughput metatranscriptome sequencing,
Journal of Biotechnology, 158(4), 248-258.
Zhang W., Li F., Nie L. (2010). Integrating multiple omics analysis for microbial biology: application
and methodologies, Microbiology 156, 287-301.