Next Article in Journal
From the Laboratory to The Vineyard—Evolution of The Measurement of Grape Composition using NIR Spectroscopy towards High-Throughput Analysis
Previous Article in Journal
Investigation of RNA Editing Sites within Bound Regions of RNA-Binding Proteins
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
High-Throughput
Volume 8
Issue 4
10.3390/ht8040020
high-throughput-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
Please note that, as of 21 September 2020, High-Throughput has been renamed to BioTech and is now published here.
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Comparative Analysis of Strategies for De Novo Transcriptome Assembly in Prokaryotes: Streptomyces clavuligerus as a Case Study

by
Carlos Caicedo-Montoya
,
Laura Pinilla
,
León F. Toro
,
Jeferyd Yepes-García
and
Rigoberto Ríos-Estepa
*
Grupo de Bioprocesos, Departamento de Ingeniería Química, Universidad de Antioquia UdeA, Calle 70 No. 52-21, Medellín 050010, Colombia
*
Author to whom correspondence should be addressed.
High-Throughput 2019, 8(4), 20; https://doi.org/10.3390/ht8040020
Submission received: 31 October 2019 / Revised: 20 November 2019 / Accepted: 23 November 2019 / Published: 30 November 2019
Browse Figures
Versions Notes

Abstract

:
The performance of software tools for de novo transcriptome assembly greatly depends on the selection of software parameters. Up to now, the development of de novo transcriptome assembly for prokaryotes has not been as remarkable as that for eukaryotes. In this contribution, Rockhopper2 was used to perform a comparative transcriptome analysis of Streptomyces clavuligerus exposed to diverse environmental conditions. The study focused on assessing the incidence of software parameters on software performance for the identification of differentially expressed genes as a final goal. For this, a statistical optimization was performed using the Transrate Assembly Score (TAS). TAS was also used for evaluating the software performance and for comparing it with related tools, e.g., Trinity. Transcriptome redundancy and completeness were also considered for this analysis. Rockhopper2 and Trinity reached a TAS value of 0.55092 and 0.58337, respectively. Trinity assembles transcriptomes with high redundancy, with 55.6% of transcripts having some duplicates. Additionally, we observed that the total number of differentially expressed genes (DEG) and their annotation greatly depends on the method used for removing redundancy and the tools used for transcript quantification. To our knowledge, this is the first work aimed at assessing de novo assembly software for prokaryotic organisms.

1. Introduction

The study of transcriptomes is essential for interpreting the genome functional elements and revealing their molecular constituents [1]. Goals for studying transcriptomes include identifying transcripts, characterizing transcript structural complexity and coding content, and more important, understanding which genes are expressed in different samples under various conditions and their levels of expression [2,3].
Traditionally, microarray analysis has been the technique used for profiling the global expression of genes. This hybridization-based technology is restricted to known genes and has a limited range of quantification. Nowadays, RNA sequencing (RNA-seq) is the most popular procedure; it consists of an experimental methodology that uses next-generation sequencing (NGS) technologies to generate DNA sequence reads derived from the entire RNA molecule to determine the primary sequence and relative abundance of each RNA [2,4].
Once sequencing is performed, the resulting reads are aligned to either a reference genome or reference transcripts, or assembled de novo without the genomic sequence, thus acquiring a complete transcription map [1]. De novo transcriptome assembly is necessary for organisms whose genomes have been neither sequenced nor annotated, e.g., for non-model organisms, when analyzing complex microbial communities, in meta-transcriptome studies, or while investigating uncultivable microorganisms [5,6,7]. Many software tools have been developed to assemble transcriptomes using the de novo strategy. The most commonly used are: Trinity [3,8], Oases [6], Bridger [9], SOAPdenovo-Trans [10], IDBA-Trans [11], SSP [12], Shannon [13], BinPacker [14] and Rockhopper2 [5].
De novo assembly is very sensitive to software parameters due to the lack of a genome to guide the assembly and the type of algorithms used which are mostly based on the De Bruijn graphs. Thus, they depend on the k-mer length [14] or on the minimum k-mer coverage [3]. Moreover, the consistency and biological relevance of the data, obtained from different sources, make it challenging to select the most accurate assembly [15,16], and the same data can generate substantially different assemblies, both within and between assembly methods, affecting the biological analysis and its conclusions [17]. In this regard, some authors have undertaken the task of evaluating the impact of different methodologies and software configurations on the quality of the assembled transcriptome [18,19,20,21,22].
Different software are available for evaluating the quality of a de novo assembly, e.g., SCAN [23], rnaQUAST [24], DETONATE [16], Transrate [17] and recently, a topology-based method called Branching Measure [25]. For the case of Transrate, it renders a Transrate Assembly Score (TAS) that allows for not only selecting the best assembler but also optimizing and tuning the assembler’s parameters; for this, it uses solely the sequenced reads and the assembly as inputs [17]. In summary, assembly assessments are essential for the evaluation of new methods, or in the combination of assemblies as part of optimization strategies [26]. Since no package outperforms others, for the same dataset, very different assemblies are obtained depending on the parameters used. Hence, it is always a good practice to optimize the software parameters used for a given organism [15,19,27,28].
Furthermore, many assemblers and methodologies were developed using eukaryotic transcriptome data. Related works have not been performed in bacteria wherein processes, structures and transcriptional mechanisms are different, having transcriptional units that often overlap, polycistronic transcripts, different promoters that may drive expression of a gene or operon under different conditions, etc., [29,30]. At the time of writing this manuscript, only two software packages had been designed specifically for bacterial transcriptome: SPARTA [31] and Rockhopper2 [5]; both of them run on a reference-based mode but, for de novo transcriptome assembly, only Rockhopper2 has been implemented.
This work aimed at performing an assembler parameter optimization; having in mind that strategies for evaluating the quality of transcriptome assembly in prokaryotes are rather scarce, we assessed the incidence of software parameters on software performance for the identification of differentially expressed genes as a final goal. Further, we performed a comparison between assemblers as well as developed approaches to assess and remove redundancy during data processing. We found that the assembly method, the redundancy removal and transcript quantification affect the quantity and the identity of deferentially expressed genes, thus modifying the biological analysis and results.

2. Materials and Methods

2.1. RNA Extraction, Library Preparation and Sequencing

Streptomyces clavuligerus (S. clavuligerus) American Type Culture Collection (ATCC) 27064 RNA samples were taken during the stationary phase of the cells, grown under favorable and unfavorable environmental conditions for clavulanic acid (CA) production. Experimental conditions were set up at 28 °C, pH 6.8 and shaking at 220 rpm. For the high CA production condition, an isolated soy-protein based medium was used [32]; for the low production condition, the chemically-defined media GSPG medium was employed [33]. RNA extraction was performed using the TRIzol® RNA Isolation Protocol [34]. The quantity of RNA was determined using a Nanodrop spectrophotometer. The RNA integrity number (RIN) was verified with the Agilent 2100 Bioanalyzer. For the cDNA library preparation, the TruSeq stranded mRNA Sample Preparation Kit (Illumina) was used. The Ilumina Hi-seq 2000 platform was utilized for RNA sequencing, and 2 x 101 bp paired-end libraries were obtained. These experiments were performed elsewhere [35] so the sequencing data was available for performing the current study. Raw RNA-seq data were deposited in the National Center for Biotechnology Information´s (NCBI) Sequence Read Archive (SRA) under accession number SAMN11046362.

2.2. Pre-Processing of Raw Data

FastQC was used for checking the quality of raw data; [36]. Subsequently, SortMeRNA [37] was used to remove ribosomal RNA. The procedure continued with a filtering of sequences by quality and an adapter trimming process, using Trimmomatic [38]. The “paired-end data” and “Truseq3” options were selected. The trimming steps were ILLUMINACLIP and SLIDINGWINDOW, both of them using the default parameters. Finally, an error correction procedure was implemented with the package BBNorm [39]. The quality of the libraries was checked with FastQC after each step in the pre-processing of raw data to ensure a successful pre-treatment of reads.

2.3. Transcriptome Assembly

Transcriptome assembly was performed using Rockhopper2 [5]. A parameter optimization was carried out through a factorial and two central composite experimental designs, developed using STATGRAPHICS Centurion XVI®. The software parameters were defined as variables; the response variable was the Transrate Assembly Score (TAS). Subsequently, the performance of Rockhopper2 was compared with that of Trinity [3,40], using TAS as response variable; Trinity was run using the default parameters.

2.4. Transcriptome Assembly Quality Evaluation

Additional metrics were also computed for analyzing the quality of the assembled transcriptomes. We evaluated the completeness of the transcriptomes based on evolutionary expectations of gene content from near-universal single-copy orthologues selected from the database OrthoDB by using the software BUSCO [41]. We built the representation of full-length reconstructed protein-coding genes, through BLASTN search by using the command analyze_blastPlus_topHit_coverage, included in Trinity. Besides, we considered the percentage of fragments that mapped to the assembled transcriptomes, handed in by Transrate. Finally, the redundancy of transcriptomes was examined using BLAST homology search and the information in “outfmt6” format was processed using a Microsoft Excel spreadsheet. Moreover, analysis of vector contamination (VecScreen [42], and chimera detection (Vsearch [43] and CD-HIT-DUP [44]), both software implemented in the Galaxy platform [45,46], were utilized for detecting chimeras in the assembled transcriptome. The results allowed us to propose two methodologies for the elimination of redundancy. First, a software that cluster sequences based on sequence similarity was used; for this, we assessed CD-HIT [44]. Second, we removed redundancy based on a BLAST homology search, which was performed using TransPS [47].

2.5. Transcript Quantification and Differential Expression Analysis

RNA-Seq by Expectation-Maximization (RSEM) [48] and Salmon [49] were used to evaluate the quantification of transcripts. Normalization and differential expression analysis was developed with the Bioconductor package EdgeR [50]. To be considered as differentially expressed (DE), genes must have a False Discovery Rate (FDR) <0.05. A gene with a fold-change larger than 2 (Log2FC >2) was considered upregulated. The dispersion parameter was set to 0.1 following the recommendations of the EdgeR manual for genetically identical organisms. A volcano plot, the application of a Bland-Altman plot (MA) plot and heat map were acquired following the scripts defined by Trinity [3]. Enrichment analysis for genes upregulated in both conditions (high and low production of CA) was performed through the R package GOseq [51]; for the visualization of the results we used REVIGO [52].

2.6. Annotation

Transdecoder, a companion of the de novo transcriptome assembler Trinity [3], was used to identify candidate coding regions within transcript sequences. Trinotate [53], powered by BLAST+ [54], HMMer [55], Signalp and TMHMM [56], was utilized for automatic functional annotation of transcriptomes. For all BLAST+ searches, a nucleotides database of twenty Streptomyces species was created, including S. clavuligerus. In addition, for BLASTX searches, all the terms associated with the Streptomyces genus were downloaded from the TrEMBL database. Finally, Trinotate uses the Swiss-Prot database to perform BLASTX searches and to obtain the gene ontology terms.

2.7. Referenced-Based Analysis

A reference-based strategy was implemented in the Galaxy platform for comparing the results of the de novo strategy for transcriptome assembly. The reference genome and the annotations for S. clavuligerus were downloaded from Ensembl Bacteria [57]. Alignments to the reference genome was performed with Bowtie 2 [58]; for transcripts quantification, Htseq-count [59] was used. Normalization and differential expression analysis followed the same technique as for the de novo strategy. Figure 1 shows the corresponding workflow chart.

3. Results

3.1. Quality Control of Raw Data

Sequencing the libraries for the high and low CA production conditions produced high quality reads (Figure S1). The FastQC report shows an adapter contamination; the GC distribution over all sequences suggested a contamination with ribosomal RNA (sharp peak in the plot); the overrepresented sequences module confirmed this result. The GC content reached 57% for both libraries. Looking for possible contamination with rRNA, reads were aligned against the SILVA [60] and RFAM [61] databases, and the hits were removed with SortMeRNA [37]. Bases that showed suboptimal quality were trimmed, and the adapter content removed with Trimmomatic (Table S1 summarizes the results of the pre-assembly processes and the quality control). In addition, an error correction process, based on the k-mer content of the reads, was performed.

3.2. Transcriptome Assembly

Rockhopper2 was initially used since it was developed specifically for bacterial transcriptomes [5]. Due to the fact that transcriptome assembly is a core stage of RNA-seq data analysis, and different assemblers produce different results for a given organism, it is always recommended to explore the parameter space to find optimized values [15,19]. Therefore, prior to any further use, the analysis began performing a parameter optimization for the software. First, a factorial experimental design was carried out (Table S2). Variables were defined as the Rockhopper2 parameters; Transrate Assembly Score (TAS) was set as the response variable. A TAS of 0.2746 was reached using the conditions set for the experiment 4 (Table S2), while for the default parameters, a TAS of 0.1331 was obtained. All parameters showed a meaningful impact on the value of TAS. The behavior of the response variable depending on the parameters value can be seen in Figure 2, wherein a TAS linear increment is observed with both an increase in “B: Minimun transcript length” and a decrease in “A: Min reads mapping to a transcript”.
Based on results from the factorial design, a central composite experimental design (CCD1) over the point with the maximum TAS (previously obtained) was proposed for continuing to improve the quality of the transcriptome assembly (Table S3). Process variables and the response variable were set as before. A TAS value of 0.3804 was reached with the conditions of Experiment 7; this is, A = 1 (Min reads mapping to a transcript), B = 160 (Minimum transcript length), C = 20 (Min count to seed a transcript) and D = 1 (Min count to extend a transcript) (Table S3). The parameter D did not show a significant effect on the TAS value (Figure 3); therefore, it was set to one and a new central composite design was established (CCD2). For this, the value of the parameter, when the maximum TAS was obtained, was set as the central point (Table S4); thus exploring values around this in an attempt for finding an optimum that contribute to a better assembly. Figure 4 shows the variation of the response variable as a function of the software parameters. The parameter B “minimum transcript length” caused the main effect on the response variable. The surface predicts a proportional TAS increase with this parameter. By default, Rockhopper2 uses 50 as a value for “minimum transcripts length”; this is a low value considering that the shorter the transcripts, the lower the probability of correspondence to real expressed transcripts.
Additional in silico experiments did not improve the TAS value, despite an increase in B: minimum transcript length. The optimum TAS value was 0.4885 (Run 17 in Table S4). A further improvement (TAS = 0.55092) was achieved by merging both libraries and assembling the transcriptome, an approach recommended by Trinity [3]. Table 1 shows the results from different metrics used for evaluating the quality of the transcriptome, assembled with Rockhopper2, both, and using the default and the optimized parameters. As observed, the quality of the assembled transcriptomes is very sensible to the software parameters:, thus, parameter optimization allowed us to reach a TAS of 0.55092; using the default parameters, the TAS was 0.19901 (merging both libraries). Additional measurements (such as the number of mapped fragments) particularly increased (75% using the default parameters and 88% using the optimized parameters). Error correction using BBnorm allowed a slight improvement in the transcriptome quality, thus rendering TAS values near 0.55709.
The transcriptome assembly was carried out in Trinity using the default parameters; the results were compared with Rockkhopper2 performance. A TAS of 0.58337 and 94% aligned fragments were observed. Table 1 shows the comparisons between the optimized Rockhopper2 and Trinity assemblers. Trinity outperforms Rockhopper2 despite the in-depth parameter optimization. According to Haas et al., Trinity reconstructs transcripts accurately with a simple and intuitive interface that requires either little or no parameter tuning [3].
Other metrics are improved by parameter optimization, e.g. “Number with Open Reading Frame (ORF)”, which indicates the transcripts that potentially encode for a protein. We calculated the mean length for the genes of the chromosome of S. clavuligerus ATCC 27064 as 343.06; the mean length calculated from the Trinity assembly is close to that from the optimized Rockhopper2.

3.3. Post-Assembly Quality Control

The analysis of the quality of the transcriptome cannot be limited to TAS, so further methodologies must be used, e.g., completeness assessment and redundancy determination. Two transcriptomes, obtained using different software solutions (optimized Rockhopper2 and Trinity), were exposed to different strategies to remove redundancy and completeness evaluation. First, redundancy was evaluated by searching for homologies in BLASTN (see Figure 5) [62]. The BLASTN search allows for identifying which transcripts that have the same hit are going to be scaffolded to a contig having a gap (duplicates without overlapping), or those that have superposition and therefore may be assembled to a larger contig (duplicates with overlapping). This redundancy of transcriptome library is inevitably existent since some transcripts, belonging to the same gene, have no overlapping regions so that they cannot be assembled together [63]. Moreover, searching homologies in BLASTN also permitted to ascertain additional errors obtained from Trinity, e.g., the same sequence assembled twice in the same transcript (self-chimeras). Though few cases were detected (6 out of 6488 assembled transcripts), this occurrence was not noticed in Rockhopper2.
In this regard, despite the fact that the TAS for Trinity surpasses that for Rockhopper2, assembly from Trinity shows larger redundancy, which needs to be removed. For this, two strategies were used: first, CD-HIT removed 115 transcripts with more than 80% identity with the transcriptome from Trinity. For the case of Rockhopper2, removal of redundancy using CD-HIT did not render results, which means that Rockhopper2 does not assemble contigs with at least 80% identity.
As a second strategy, we used scaffolding of contigs using TransPS. The methodology for TransPS deals with the search for homologies in BLASTX; based on such homology, scaffolding of contigs that codify for the same sequence is completed. In this manner, redundancy is removed, and, in some cases, longer transcripts are attained.
Consequently, three transcriptomes were achieved: Trinity.CD-HIT, Trinity.TransPS and Rockhopper.TransPS, with transcript numbers of 6373, 3285 and 2481, and TAS values of 0.6122, 0.5303 and 0.5405, respectively (see Table 2). The percentage of reads alignment to the assembled transcriptome was 94%, 83%, and 76%, respectively. For the removal of redundancy, TAS increases with CD-HIT and diminishes with TransPS.
A further methodology used for testing transcriptome quality and completeness was Benchmarking Universal Single-copy Orthologs (BUSCO) (see Figure 6). As a summary, Trinity assembles transcriptomes with a larger number of fragmented transcripts, whereas Rockhopper2 presents a larger number of missing contigs. For both Trinity and Rockhopper2, the number of complete contigs was similar, though Trinity showed a tendency to cause more duplicates, which is consistent with results in Figure 5.
Finally, full-length transcripts were determined (see Figure 7). Ideally, all transcripts must have a percentage of the hit’s length equal to 100%, but due to the uneven coverage of the libraries, not all of the transcripts are assembled completely. Besides these metrics, an analysis of the assembled transcriptome was performed using VecScreen, for removing vector contamination. Nevertheless, for the case of the transcriptome assembled using Trinity, only 10 transcripts out of 6488 were detected with vector contamination, which represents only 0.15% of the total assembled transcripts. Also, Vsearch Chimera detection and CD-HIT-DUP were utilized in order to detect chimeras in the assembled transcriptome. No transcripts were categorized as chimera. Hence, an alternative methodology was proposed; it refers to aligning in BLASTN the transcripts with a score given by Transrate, sCseg less than 0.1, since a transcript with a low sCseg is possibly a chimera. Only three chimeras were found.

3.4. Transcript Quantification and Differential Expression Analysis

The resulting transcriptomes (Trinity.CD-HIT, Trinity.TransPS and Rockhopper.TransPS) were treated with the transcript quantification software RSEM and Salmon to compare an alignment-based method (RSEM) against an alignment-free method (Salmon). The first approach was to know how the software type for transcript quantification affects the number of genes classified as differentially expressed. Figure 8A,B show the differentially expressed genes (DEG) based on the software used. Results also include assembling outcomes using the S. clavuligerus genome as reference; from this, 97% of reads did align properly and 260 genes were detected as differentially expressed.
For the case of removal of redundancy using CD-HIT and quantification by both methods, we found transcripts with at least a duplicate in the group of DEG (25 and 26, using RSEM and Salmon, respectively). Once this software is used, problems related with redundancy are dragged all the way until the end of the DEG analysis. Results for the differential expression analysis for the Rockhopper.TransPS transcriptome are presented in the Supplementary Material (Figure S2A,B). As observed, a high proportion of genes detected as differentially expressed are upregulated in the high CA production condition. The gene differentially expressed with the greatest LogFC, i.e., the most upregulated, was a neutral-zinc metalloprotease, highly expressed jointly with a large group of proteases in the high CA production condition.

3.5. Annotation and Gene Ontology

Figure 9 shows a summary for the GO results from the assembled transcriptomes. In Figure 9A, processes associated with CA biosynthesis, proline biosynthesis, cell communication, proteolysis and protein transport are enriched in this plot for the Trinity.TransPS transcriptome, while, for Figure 9B, regulation of nitrogen utilization, lipid transport and growth of symbiont in the host cell are enriched for the Rockhopper.TransPS transcriptome. Therefore, an analysis based on a single transcriptome would consider biological scenarios that the other transcriptome might not, thus biasing the analysis, e.g., it is expected that the methodology would provide nodes linked to secondary metabolism, specifically, CA biosynthesis; however, from one method (Rockhopper.TransPS), such a node was not annotated. Accordingly, it could be inferred that genes associated with CA production are not expressed in the high production condition when indeed they are (See annotations in Figure 9A and reference [35]).

4. Discussion

In this work, a parameter optimization of the assembler software Rockhopper2 was undertaken, and the quality of the transcriptome obtained was analyzed using the Transrate Assembly Score (TAS).
The quality of the assembled transcriptomes, using Rockhopper2, is very sensible to the software parameters. Parameter optimization allowed reaching a satisfactory TAS of 0.55092; using the Rockhopper2 default parameters, the TAS was 0.19901. Indeed, this is a reasonable improvement if we consider that Smith and Unna calculated TAS values for 155 assemblies in the range of 0.001 and 0.52. A recent work of Holzer and Marz used different metrics, including TAS, to analyze ten assemblers using different RNA-seq data from diverse sources. The authors reached high TAS values for Candida albicans (0.54236) and Homo sapiens + Evola virus (EBOV) (0.5475) [64]. Values attained by these authors resemble the values reached in the present work for the TAS.
Overall, these results indicate that parameter optimization is needed and highly recommended for better de novo assembly. This work revealed that all parameters in Rockhopper2 define the quality of the assembled transcriptomes in terms of TAS. The parameter “min count to seed a transcript” is responsible for the starting of assembling a candidate transcript if at least the specified number of k-mers are present in the reads. Meanwhile, “min count to extend a transcript” deals with extending a candidate transcript from a node in the De Bruijn graph if at least the specified number of k-mers in the reads correspond to the subsequent node. Further, a candidate transcript is considered as high quality transcript if it has surpassed the threshold determined by “minimun transcript length” and by “min reads mapping to a transcript” [5]. In this work, the optimization tended to decrease the parameters “min reads mapping to a transcript”, “min count to seed a transcript” and “min count to extend a transcript”, which might be caused by the low coverage for the organism under study. Thus, many transcripts present in the sample were not initially assembled since there were not enough reads, and consequently the k-mer frequency was too low to fill the conditions required for the software parameters. By decreasing these parameters, many more contigs can be assembled and more reads can be used for the assembling process, thus increasing quality.
The performance of Trinity has been analyzed elsewhere [19,20,21,22,25,28,65,66,67,68,69,70]. In general, Trinity assembles more contiguous transcripts; it has a good performance at assembling conserved genes across a large set [71]; it also has the ability to detect alternative splice isoforms and does not require parameter tuning [3]. All this in contrast to its long time and high Random Acces Memory (RAM) required for assembly, which make it necessary to have high computational resources. Nowadays, it is available at some platforms e.g., Galaxy [45] or TRUFA [72].
According to a previous report, Trinity produced few chimeras across all single k-mer assemblies of the Arabidopsis thaliana transcriptome [70]. Conversely, Yang and Smith reported that Trinity produces large amounts of trans self-chimeras [20]. In this work, six self-chimeras out of 6488 transcripts were found. No additional chimeras were found using Vsearch Quimera Detection [43] and CD-HIT-DUP [44], and three were found using the proposed alternative.
In terms of TAS, the performance of Rockhopper2 and Trinity was quite similar, despite the in-depth parameter optimization in Rockhopper2. Trinity assembles almost double the transcripts compared with Rockhopper2; 94% of the reads align to the assembled transcriptome, which is quite similar to the reference-based approach (97%), yet the mean length is shorter. A high percentage of reads mapping back to the transcriptome assembly is desirable for accurate differential gene expression analysis because more reads mapped back to the assembly will result in increased statistical power for performing these analyses [21].
Redundant contigs represent highly similar sequences corresponding to the same reference. Huang et al. reported that Trinity assemblies contain more redundancy relative to assemblies generated by the genome-guided and TransPS methods. The same authors inferred this is in part due to the fact that RNA-Seq was performed on mRNAs from tissue samples that were pooled from multiple individuals [21]. In this work, the high redundancy in the Trinity assembly might be related to its ability to detect isoforms in eukaryotic genes. Therefore, highly similar transcripts that could codify for the same gene in bacteria are designated as different transcripts because of the underlying software design; yet, it could be related to intra-reads error which might lead to this situation.
Due to its scaffolding process, TransPS increases mean length both in Trinity and Rockhopper2 assemblies. Moreover, the treatment with TransPS allows for retrieving almost double the full-length transcripts assembled than CD-HIT. Therefore, the post-assembly scaffolding process carried out by TransPS is highly recommended considering its benefits in terms of redundancy reduction and full-length transcripts recovered.
Concerning the gene ontology terms annotated, both Trinity.CD-HIT and Trinity.TransPS share a high quantity of GO terms among the DEG, depending on the quantification method used, with 88% and 91% of the terms in common, respectively. The Rockhopper.TransPS transcriptome only reaches a 76% of GO terms in common for both quantification methods (Figure S3). The previous results may be related to the quantification algorithm of Salmon that uses quasi-mapping instead of traditional alignments. Thus, sometimes it could provide superior accuracy by being more robust to errors in the read or genomic variation from the reference sequence [73]. In contrast, RSEM needs a previous alignment using either bowtie [74] or bowtie 2 [58]; this could discard some information depending on the parameter used in the aligner, a topic not explored in this work. The type of data from a bacterial transcriptome where no isoforms are produced and the post-assembly procedures may explain the differences between the quantities of DEG, detected after the different method of transcript quantification. The quantity of genes considered as differentially expressed continues to be greater in the reference-based mode than in de novo; yet, de novo allows for detecting genes that the reference-based does not. This could be related to transcript fragmentation and incomplete and incorrect annotation, whereby we suggest that de novo assembly is beneficial even when a reference genome is available [70].
As a first approach, this study highlights the importance of a deep analysis of the parameters of the assembler software, and the post-assembly methodologies. Consequently, we believe that a deeper evaluation of strategies recommended by other authors, which were successful for improving the quality of assembled transcriptomes from short-reads, are required in prokaryotic transcriptomes, e.g., the use of a deeper sequencing, to merge the results from different assemblies and use a genome-guided de novo assembly. Certainly, the most interesting result will be the different annotations reached depending on the methodology; therefore, the biological conclusions will greatly depend on the selected method.

5. Conclusions

Transcriptome analysis has been mostly dedicated to eukaryotic organisms. As far as we know, this is the first work aimed at assessing software for prokaryotic organisms such as Rockhopper2. This study highlights the importance of a parameter optimization and analysis, primarily when a de novo assembly is carried out.
Previous studies reported that Rockhopper2 is superior to Trinity for assembling bacterial transcriptomes [5]; however, our findings contradict this report since the quality of the transcriptome assembled with Trinity surpassed the quality of the transcriptome assembled with Rockhopper2, even after the parameter optimization study; nonetheless, the quality was quite similar.
The analysis cannot be limited to TAS; further methodologies must be used such as completeness and redundancy. Thus, a wider panorama about the quality of the assembled transcriptome is achieved. Similar conclusions were reached by Honaas et al. [22], who concluded that the quality of de novo transcriptome assemblies is best assessed through consideration of a combination of metrics. CD-HIT does not provide proper results when removing redundancy, since various transcripts with duplicates are detected as differentially expressed. For this, a methodology based on homology searching using BLAST, such as TransPS, is suggested.
Finally, appropriate techniques for transcriptome analysis are highly recommended since biological outcomes will rely on the selected method. Thus, different annotations, depending on the method for removal of redundancy and transcript quantification, may be accomplished.

Supplementary Materials

The following are available online at https://www.mdpi.com/2571-5135/8/4/20/s1, Figure S1: Quality control, Figure S2: MA and Volcano plots, Figure S3: Gene ontology terms shared between transcriptomes evaluated, Table S1: Basic statistics of the pre-assembly stages, Table S2: Factorial design for the Rockhopper2 parameters in de novo mode, Table S3: Central composite design 1 for the Rockhopper2 parameters in de novo mode, Table S4: Central composite design 2 for the Rockhopper2 parameters in de novo mode.

Author Contributions

C.C.-M.: conceived the study, designed the experiments, performed all the bioinformatics analysis and drafted the manuscript. L.P., L.F.T. and J.Y.-G.: executed the RNA extraction, library preparation and RNA sequencing. R.R.-E.: supervised the research work, interpreted the results, corrected and wrote the manuscript and serve as corresponding author. All authors have read and approved the final manuscript.

Funding

This work was supported by Departamento Administrativo de Ciencia, Tecnología e Innovación–COLCIENCIAS-Colombia (Grant Number 111566945929).

Acknowledgments

The authors thank Mauricio Rodriguez from Universidad de Antioquia, CO, for helpful comments and discussions. L.F.T and L.P. thank COLCIENCIAS-Colombia for scholarships.

Conflicts of Interest

The authors declare no conflict of interest regarding the publication of this paper.

References

  1. Wang, Z.; Gerstein, M.; Snyder, M. RNA-Seq: A revolutionary tool for transcriptomics. Nat. Rev. Genet. 2009, 10, 57–63. [Google Scholar] [CrossRef]
  2. Martin, J.A.; Wang, Z. Next-generation transcriptome assembly. Nat. Rev. Genet. 2011, 12, 671–682. [Google Scholar] [CrossRef] [PubMed]
  3. Haas, B.J.; Papanicolaou, A.; Yassour, M.; Grabherr, M.; Blood, P.D.; Bowden, J.; Couger, M.B.; Eccles, D.; Li, B.; Lieber, M.; et al. De novo transcript sequence reconstruction from RNA-Seq: Reference generation and analysis with Trinity. Nat. Protoc. 2013, 8, 1–43. [Google Scholar] [CrossRef] [PubMed]
  4. van Verk, M.C.; Hickman, R.; Pieterse, C.M.J.; van Wees, S.C.M. RNA-Seq: Revelation of the messengers. Trends Plant Sci. 2013, 18, 175–179. [Google Scholar] [CrossRef] [PubMed]
  5. Tjaden, B. De novo assembly of bacterial transcriptomes from RNA-seq data. Genome Biol. 2015, 16. [Google Scholar] [CrossRef]
  6. Schulz, M.H.; Zerbino, D.R.; Vingron, M.; Birney, E. Oases: Robust de novo RNA-seq assembly across the dynamic range of expression levels. Bioinformatics 2012, 28, 1086–1092. [Google Scholar] [CrossRef]
  7. Birol, I.; Jackman, S.D.; Nielsen, C.B.; Qian, J.Q.; Varhol, R.; Stazyk, G.; Morin, R.D.; Zhao, Y.; Hirst, M.; Schein, J.E.; et al. De novo transcriptome assembly with ABySS. Bioinformatics 2009, 25, 2872–2877. [Google Scholar] [CrossRef]
  8. Grabherr, M.G.; Haas, B.J.; Yassour, M.; Levin, J.Z.; Thompson, D.A.; Amit, I.; Adiconis, X.; Fan, L.; Raychowdhury, R.; Zeng, Q.; et al. Full-length transcriptome assembly from RNA-Seq data without a reference genome. Nat. Biotechnol. 2011, 29, 644–652. [Google Scholar] [CrossRef]
  9. Chang, Z.; Li, G.; Liu, J.; Zhang, Y.; Ashby, C.; Liu, D.; Cramer, C.L.; Huang, X. Bridger: A new framework for de novo transcriptome assembly using RNA-seq data. Genome Biol. 2015, 16, 30. [Google Scholar] [CrossRef]
  10. Xie, Y.; Wu, G.; Tang, J.; Luo, R.; Patterson, J.; Liu, S.; Huang, W.; He, G.; Gu, S.; Li, S.; et al. SOAPdenovo-Trans: De novo transcriptome assembly with short RNA-Seq reads. Bioinformatics 2014, 30, 1660–1666. [Google Scholar] [CrossRef]
  11. Peng, Y.; Leung, H.C.M.; Yiu, S.M.; Lv, M.J.; Zhu, X.G.; Chin, F.Y.L. IDBA-tran: A more robust de novo de Bruijn graph assembler for transcriptomes with uneven expression levels. Bioinformatics 2013, 29, 326–334. [Google Scholar] [CrossRef] [PubMed]
  12. Safikhani, Z.; Sadeghi, M.; Pezeshk, H.; Eslahchi, C. SSP: An interval integer linear programming for de novo transcriptome assembly and isoform discovery of RNA-seq reads. Genomics 2013, 102, 507–514. [Google Scholar] [CrossRef] [PubMed]
  13. Kannan, S.; Hui, J.; Mazooji, K. Shannon: An Information-Optimal de Novo RNA-Seq Assembler. bioRxiv 2016. [Google Scholar] [CrossRef]
  14. Liu, J.; Li, G.; Chang, Z.; Yu, T.; Liu, B.; McMullen, R.; Chen, P.; Huang, X. BinPacker: Packing-Based De Novo Transcriptome Assembly from RNA-seq Data. PLoS Comput. Biol. 2016, 12, 1–15. [Google Scholar] [CrossRef]
  15. Conesa, A.; Madrigal, P.; Tarazona, S.; Gomez-Cabrero, D.; Cervera, A.; McPherson, A.; Szcześniak, M.W.; Gaffney, D.J.; Elo, L.L.; Zhang, X.; et al. A survey of best practices for RNA-seq data analysis. Genome Biol. 2016, 17, 13. [Google Scholar] [CrossRef]
  16. Li, B.; Fillmore, N.; Bai, Y.; Collins, M.; Thomson, J.A.; Stewart, R.; Dewey, C.N. Evaluation of de novo transcriptome assemblies from RNA-Seq data. Genome Biol. 2014, 15, 553. [Google Scholar] [CrossRef]
  17. Smith-Unna, R.; Boursnell, C.; Patro, R.; Hibberd, J.M.; Kelly, S. TransRate reference-free quality assessment of de novo transcriptome assemblies. Genome Res. 2016, 26, 1134–1144. [Google Scholar] [CrossRef]
  18. Surget-Groba, Y.; Montoya-Burgos, J. Optimization of de novo transcriptome assembly from next-generation sequencing data. Genome Res. 2010, 20, 1432–1440. [Google Scholar] [CrossRef]
  19. Zhao, Q.-Y.; Wang, Y.; Kong, Y.-M.; Luo, D.; Li, X.; Hao, P. Optimizing de novo transcriptome assembly from short-read RNA-Seq data: A comparative study. BMC Bioinform. 2011, 12, S2. [Google Scholar] [CrossRef]
  20. Yang, Y.; Smith, S.A. Optimizing de novo assembly of short-read RNA-seq data for phylogenomics. BMC Genom. 2013, 14, 328. [Google Scholar] [CrossRef]
  21. Huang, X.; Chen, X.-G.; Armbruster, P.A. Comparative performance of transcriptome assembly methods for non-model organisms. BMC Genom. 2016, 17, 523. [Google Scholar] [CrossRef] [PubMed]
  22. Honaas, L.A.; Wafula, E.K.; Wickett, N.J.; Der, J.P.; Zhang, Y.; Edger, P.P.; Altman, N.S.; Pires, J.C.; Leebens-Mack, J.H.; DePamphilis, C.W. Selecting superior de novo transcriptome assemblies: Lessons learned by leveraging the best plant genome. PLoS ONE 2016, 11, e0146062. [Google Scholar] [CrossRef] [PubMed]
  23. Misner, I.; Bicep, C.; Lopez, P.; Halary, S.; Bapteste, E.; Lane, C.E. Sequence comparative analysis using networks: Software for evaluating de novo transcript assembly from next-generation sequencing. Mol. Biol. Evol. 2013, 30, 1975–1986. [Google Scholar] [CrossRef] [PubMed]
  24. Bushmanova, E.; Antipov, D.; Lapidus, A.; Suvorov, V.; Prjibelski, A.D. RnaQUAST: A quality assessment tool for de novo transcriptome assemblies. Bioinformatics 2016, 32, 2210–2212. [Google Scholar] [CrossRef]
  25. Lima, L.; Sinaimeri, B.; Sacomoto, G.; Lopez-Maestre, H.; Marchet, C.; Miele, V.; Sagot, M.F.; Lacroix, V. Playing hide and seek with repeats in local and global de novo transcriptome assembly of short RNA-seq reads. Algorithms Mol. Biol. 2017, 12, 2. [Google Scholar] [CrossRef]
  26. Moreton, J.; Izquierdo, A.; Emes, R.D. Assembly, assessment, and availability of De novo generated eukaryotic transcriptomes. Front. Genet. 2016, 6, 1–9. [Google Scholar] [CrossRef] [PubMed]
  27. Cresko Lab. RNA-Seqlopedia. Available online: http://rnaseq.uoregon.edu/ (accessed on 22 August 2016).
  28. Cabau, C.; Escudié, F.; Djari, A.; Guiguen, Y.; Bobe, J.; Klopp, C. Compacting and correcting Trinity and Oases RNA-Seq de novo assemblies. PeerJ 2017, 5, e2988. [Google Scholar] [CrossRef] [PubMed]
  29. McClure, R.; Balasubramanian, D.; Sun, Y.; Bobrovskyy, M.; Sumby, P.; Genco, C.A.; Vanderpool, C.K.; Tjaden, B. Computational analysis of bacterial RNA-Seq data. Nucleic Acids Res. 2013, 41, e140. [Google Scholar] [CrossRef] [PubMed]
  30. James, K.; Cockell, S.J.; Zenkin, N. Deep sequencing approaches for the analysis of prokaryotic transcriptional boundaries and dynamics. Methods 2017, 120, 76–84. [Google Scholar] [CrossRef]
  31. Johnson, B.K.; Scholz, M.B.; Teal, T.K.; Abramovitch, R.B. SPARTA: Simple Program for Automated reference-based bacterial RNA-seq Transcriptome Analysis. BMC Bioinform. 2016, 17, 66. [Google Scholar] [CrossRef]
  32. Pinilla, L.; Toro, L.F.; Avignone-Rossa, C.; Peñuela, M.; Rios-Estepa, R. Streptomyces clavuligerus strain selection for clavulanic acid biosynthesis: A study based on culture composition effects and statistical analysis. DYNA 2018, 85, 111–118. [Google Scholar] [CrossRef]
  33. Romero, J.; Liras, P.; Martin, J.F. Dissociation of cephamycin and clavulanic acid biosynthesis in Streptomyces clavuligerus. Appl. Microbiol. Biotechnol. 1984, 20, 318–325. [Google Scholar] [CrossRef]
  34. Chomczynski, P.; Mackey, K. Modification of the TRI reagent procedure for isolation of RNA from polysaccharide- and proteoglycan-rich sources. Biotechniques 1995, 19, 942–945. [Google Scholar] [PubMed]
  35. Pinilla, L.; Toro, L.F.; Laing, E.; Alzate, J.F.; Ríos-Estepa, R. Comparative Transcriptome Analysis of Streptomyces Clavuligerus in Response to Favorable and Restrictive Nutritional Conditions. Antibiotics 2019, 8, 96. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  36. Andrews, S. FastQC. Available online: http://www.bioinformatics.babraham.ac.uk/projects/fastqc/ (accessed on 1 November 2015).
  37. Kopylova, E.; Noé, L.; Touzet, H. SortMeRNA: Fast and accurate filtering of ribosomal RNAs in metatranscriptomic data. Bioinformatics 2012, 28, 3211–3217. [Google Scholar] [CrossRef]
  38. Bolger, B.; Lohse, A.M.; Usadel, M. Trimmomatic: A flexible trimmer for Illumina sequence data. Bioinformatics 2014, 30, 2114. [Google Scholar] [CrossRef] [Green Version]
  39. Bushnell, B. BBMap: A Fast, Accurate, Splice-Aware Aligner; Lawrence Berkeley National Lab.(LBNL): Berkeley, CA, USA, 2014. [Google Scholar]
  40. Grabherr, M.G.; Haas, B.J.; Yassour, M.; Levin, J.Z.; Thompson, D.A.; Amit, I.; Adiconis, X.; Fan, L.; Raychowdhury, R.; Zeng, Q.; et al. Trinity: Reconstructing a full-length transcriptome without a genome from RNA-Seq data. Nat. Biotechnol. 2013, 29, 644–652. [Google Scholar] [CrossRef] [Green Version]
  41. Simão, F.A.; Waterhouse, R.M.; Ioannidis, P.; Kriventseva, E.V.; Zdobnov, E.M. BUSCO: Assessing genome assembly and annotation completeness with single-copy orthologs. Bioinformatics 2015, 31, 3210–3212. [Google Scholar] [CrossRef] [Green Version]
  42. VecScreen: Screen a Sequence for Vector Contamination. Available online: https://www.ncbi.nlm.nih.gov/tools/vecscreen/ (accessed on 17 May 2017).
  43. Rognes, T.; Flouri, T.; Nichols, B.; Quince, C.; Mahé, F. VSEARCH: A versatile open source tool for metagenomics. PeerJ 2016, 4, e2584. [Google Scholar] [CrossRef]
  44. Fu, L.; Niu, B.; Zhu, Z.; Wu, S.; Li, W. CD-HIT: Accelerated for clustering the next-generation sequencing data. Bioinformatics 2012, 28, 3150–3152. [Google Scholar] [CrossRef]
  45. Goecks, J.; Nekrutenko, A.; Taylor, J. Galaxy: A comprehensive approach for supporting accessible, reproducible, and transparent computational research in the life sciences. Genome Biol. 2010, 11, R86. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  46. Hillman-Jackson, J.; Clements, D.; Blankenberg, D.; Taylor, J.; Nekrutenko, A.; The Galaxy Team. Using galaxy to perform large-scale interactive data analyses. Curr. Protoc. Bioinform. 2012, 38. [Google Scholar] [CrossRef] [PubMed]
  47. Liu, M.; Adelman, Z.N.; Myles, K.M.; Zhang, L. A Transcriptome Post-Scaffolding Method for Assembling High Quality Contigs. Comput. Biol. J. 2014, 2014, 961823. [Google Scholar] [CrossRef] [PubMed]
  48. Li, B.; Dewey, C.N. RSEM: Accurate transcript quantification from RNA-Seq data with or without a reference genome. BMC Bioinform. 2011, 12, 323. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  49. Patro, R.; Duggal, G.; Love, M.I.; Irizarry, R.A.; Kingsford, C. Salmon provides accurate, fast, and bias-aware transcript expression. Nat. Methods 2017, 14, 417–419. [Google Scholar] [CrossRef] [Green Version]
  50. Robinson, M.D.; McCarthy, D.J.; Smyth, G.K. edgeR: A Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 2009, 26, 139–140. [Google Scholar] [CrossRef] [Green Version]
  51. Young, M.D.; Wakefield, M.J.; Smyth, G.K.; Oshlack, A. Gene ontology analysis for RNA-seq: Accounting for selection bias. Genome Biol. 2010, 11, R14. [Google Scholar] [CrossRef] [Green Version]
  52. Supek, F.; Bošnjak, M.; Škunca, N.; Šmuc, T. Revigo summarizes and visualizes long lists of gene ontology terms. PLoS ONE 2011, 6, e21800. [Google Scholar] [CrossRef] [Green Version]
  53. Bryant, D.M.; Johnson, K.; DiTommaso, T.; Tickle, T.; Couger, M.B.; Payzin-Dogru, D.; Lee, T.J.; Leigh, N.D.; Kuo, T.H.; Davis, F.G.; et al. A Tissue-Mapped Axolotl De Novo Transcriptome Enables Identification of Limb Regeneration Factors. Cell Rep. 2017, 18, 762–776. [Google Scholar] [CrossRef] [Green Version]
  54. Camacho, C.; Coulouris, G.; Avagyan, V.; Ma, N.; Papadopoulos, J.; Bealer, K. BLAST plus: Architecture and applications. BMC Bioinform. 2009, 10, 1. [Google Scholar] [CrossRef] [Green Version]
  55. Eddy, S.R. Accelerated profile HMM searches. PLoS Comput. Biol. 2011, 7, e1002195. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  56. Krogh, A.; Larsson, B.; von Heijne, G.; Sonnhammer, E.L. Predicting transmembrane protein topology with a hidden markov model: Application to complete genomes. J. Mol. Biol. 2001, 305, 567–580. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  57. Kersey, P.J.; Allen, J.E.; Allot, A.; Barba, M.; Boddu, S.; Bolt, B.J.; Carvalho-Silva, D.; Christensen, M.; Davis, P.; Grabmueller, C.; et al. Ensembl Genomes 2018: An integrated omics infrastructure for non-vertebrate species. Nucleic Acids Res. 2018, 46, D802–D808. [Google Scholar] [CrossRef] [PubMed]
  58. Langmead, B.; Salzberg, S.L. Fast gapped-read alignment with Bowtie 2. Nat. Methods 2013, 9, 357–359. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  59. Anders, S.; Pyl, P.T.; Huber, W. HTSeq-A Python framework to work with high-throughput sequencing data. Bioinformatics 2015, 31, 166–169. [Google Scholar] [CrossRef] [PubMed]
  60. Quast, C.; Pruesse, E.; Yilmaz, P.; Gerken, J.; Schweer, T.; Yarza, P.; Peplies, J.; Glöckner, F.O. The SILVA ribosomal RNA gene database project: Improved data processing and web-based tools. Nucleic Acids Res. 2013, 41, 590–596. [Google Scholar] [CrossRef]
  61. Griffiths-Jones, S.; Bateman, A.; Marshall, M.; Khanna, A.; Eddy, S.R. Rfam: An RNA family database. Nucleic Acids Res. 2003, 31, 439–441. [Google Scholar] [CrossRef] [Green Version]
  62. Ono, H.; Ishii, K.; Kozaki, T.; Ogiwara, I.; Kanekatsu, M.; Yamada, T. Removal of redundant contigs from de novo RNA-Seq assemblies via homology search improves accurate detection of differentially expressed genes. BMC Genom. 2015, 16, 1031. [Google Scholar] [CrossRef] [Green Version]
  63. Song, H.; Yu, Z.-L.; Sun, L.-N.; Gao, Y.; Zhang, T.; Wang, H.-Y. De novo transcriptome sequencing and analysis of Rapana venosa from six different developmental stages using Hi-seq 2500. Comp. Biochem. Physiol. Part D. Genom. Proteom. 2016, 17, 48–57. [Google Scholar] [CrossRef]
  64. Holzerand, M.; Marz, M. De novo transcriptome assembly: A comprehensive cross-species comparison of short-read RNA-Seq assemblers. GigaScience 2019, 8, 1–16. [Google Scholar]
  65. Mbandi, S.K.; Hesse, U.; van Heusden, P.; Christoffels, A. Inferring bona fide transfrags in RNA-Seq derived-transcriptome assemblies of non-model organisms. BMC Bioinform. 2015, 16, 58. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  66. He, B.; Zhao, S.; Chen, Y.; Cao, Q.; Wei, C.; Cheng, X.; Zhang, Y. Optimal assembly strategies of transcriptome related to ploidies of eukaryotic organisms. BMC Genom. 2015, 16, 65. [Google Scholar] [CrossRef] [Green Version]
  67. Lu, B.X.; Zeng, Z.B.; Shi, T.L. Comparative study of de novo assembly and genome-guided assembly strategies for transcriptome reconstruction based on RNA-Seq. Sci. China Life Sci. 2013, 56, 143–155. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  68. Amin, S.; Prentis, P.J.; Gilding, E.K.; Pavasovic, A. Assembly and annotation of a non-model gastropod (Nerita melanotragus) transcriptome: A comparison of De novo assemblers. BMC Res. Notes 2014, 7, 488. [Google Scholar] [CrossRef] [Green Version]
  69. Vijay, N.; Poelstra, J.W.; Künstner, A.; Wolf, J.B.W. Challenges and strategies in transcriptome assembly and differential gene expression quantification. A comprehensive in silico assessment of RNA-seq experiments. Mol. Ecol. 2013, 22, 620–634. [Google Scholar] [CrossRef] [PubMed]
  70. Wang, S.; Gribskov, M. Comprehensive evaluation of de novo transcriptome assembly programs and their effects on differential gene expression analysis. Bioinformatics 2016, 33, 27–323. [Google Scholar]
  71. Francis, W.R.; Christianson, L.M.; Kiko, R.; Powers, M.L.; Shaner, N.C.; Haddock, S.H.D. A comparison across non-model animals suggests an optimal sequencing depth for de novo transcriptome assembly. BMC Genom. 2013, 14, 167. [Google Scholar] [CrossRef] [Green Version]
  72. Kornobis, E.; Cabellos, L.; Aguilar, F.; Frías-López, C.; Rozas, J.; Marco, J.; Zardoya, R. TRUFA: A User-Friendly Web Server for de novo RNA-seq Analysis Using Cluster Computing. Evol. Bioinform. 2015, 11, 97–104. [Google Scholar] [CrossRef] [Green Version]
  73. Patro, R.; Duggal, G.; Love, M.; Irizarry, R.; Kingsford, C. Salmon’s Documentation. 2017. Available online: http://salmon.readthedocs.io/en/latest/ (accessed on 25 October 2017).
  74. Langmead, B. Alignment short sequencing reads with Bowtie. Curr. Protoc. Bioinform. 2010, 32, 11.7.1–11.7.14. [Google Scholar] [CrossRef]
High throughput 08 00020 g001 550
Figure 1. Workflow for the bioinformatics analysis implemented in this study.
Figure 1. Workflow for the bioinformatics analysis implemented in this study.
High throughput 08 00020 g001
High throughput 08 00020 g002 550
Figure 2. Response surface for the factorial experimental design aimed at evaluating Rockhopper2 parameters. Min count to seed a transcript = 50.0; Min count to extent a transcript = 5.0.
Figure 2. Response surface for the factorial experimental design aimed at evaluating Rockhopper2 parameters. Min count to seed a transcript = 50.0; Min count to extent a transcript = 5.0.
High throughput 08 00020 g002
High throughput 08 00020 g003 550
Figure 3. Central composite experimental design for the parameters of Rockhopper2, in the de novo mode. (A): Response surface. Min count to seed a transcript = 90.0; Min count to extent a transcript = 2.0. (B): Pareto chart for Transrate assembly score (TAS). A: Min reads mapping to a transcript; B: Minimum transcript length; C: Min count to seed a transcript. AC and AB are interaction between the previously defined factors
Figure 3. Central composite experimental design for the parameters of Rockhopper2, in the de novo mode. (A): Response surface. Min count to seed a transcript = 90.0; Min count to extent a transcript = 2.0. (B): Pareto chart for Transrate assembly score (TAS). A: Min reads mapping to a transcript; B: Minimum transcript length; C: Min count to seed a transcript. AC and AB are interaction between the previously defined factors
High throughput 08 00020 g003
High throughput 08 00020 g004 550
Figure 4. Estimated response surface for the Central Composite Experimental Design2. Min count to seed a transcript = 10.0.
Figure 4. Estimated response surface for the Central Composite Experimental Design2. Min count to seed a transcript = 10.0.
High throughput 08 00020 g004
High throughput 08 00020 g005 550
Figure 5. Post-assembly quality control. BLAST homology search. BH: BLAST Hits. SOD: Subjects with a least One Duplicate. DWoO: Duplicates WithOut Overlapping. DWO: Duplicates With Overlapping.
Figure 5. Post-assembly quality control. BLAST homology search. BH: BLAST Hits. SOD: Subjects with a least One Duplicate. DWoO: Duplicates WithOut Overlapping. DWO: Duplicates With Overlapping.
High throughput 08 00020 g005
High throughput 08 00020 g006 550
Figure 6. Benchmarking Universal Single-copy Orthologs (BUSCO) assessment results.
Figure 6. Benchmarking Universal Single-copy Orthologs (BUSCO) assessment results.
High throughput 08 00020 g006
High throughput 08 00020 g007 550
Figure 7. Full-length transcripts assembled retrieved from two assembly and two redundancy removal methods.
Figure 7. Full-length transcripts assembled retrieved from two assembly and two redundancy removal methods.
High throughput 08 00020 g007
High throughput 08 00020 g008 550
Figure 8. Differentially expressed gene analysis. (A): Venn diagram for differentially expressed genes (DEG) obtained using RNA-Seq by Expectation-Maximization (RSEM). (B): Venn diagram for DEG obtained using Salmon.
Figure 8. Differentially expressed gene analysis. (A): Venn diagram for differentially expressed genes (DEG) obtained using RNA-Seq by Expectation-Maximization (RSEM). (B): Venn diagram for DEG obtained using Salmon.
High throughput 08 00020 g008
High throughput 08 00020 g009 550
Figure 9. Gene ontology (GO) for biological processes (A): GO for upregulated genes related to clavulanic acid biosynthesis, using Trinity.TransPS (quantification method: Salmon). (B): GO for upregulated genes related with clavulanic acid biosynthesis, using Rockhopper.TransPS (quantification method: Salmon). Figures were obtained using the software REVIGO [52]. Highlighted with ----- are the nodes that differ from both annotations.
Figure 9. Gene ontology (GO) for biological processes (A): GO for upregulated genes related to clavulanic acid biosynthesis, using Trinity.TransPS (quantification method: Salmon). (B): GO for upregulated genes related with clavulanic acid biosynthesis, using Rockhopper.TransPS (quantification method: Salmon). Figures were obtained using the software REVIGO [52]. Highlighted with ----- are the nodes that differ from both annotations.
High throughput 08 00020 g009
Table
Table 1. Comparison of different metrics for the assembled transcriptome through distinctive software.
Table 1. Comparison of different metrics for the assembled transcriptome through distinctive software.
Quality MetricRockhopper2 Default ParametersRockhopper2 Optimized ParametersTrinity Default Parameters
Number of Sequences112934596488
Mean Length243.13604.088483.83
Number with ORF16515281919
Fragments638,046638,046638,046
Mapped478,534 (75%)571,882(89%)602,400(94%)
TAS0.190010.550920.58337
Table
Table 2. Quality control of the assembled transcriptomes subjected to different strategies for removing redundancy.
Table 2. Quality control of the assembled transcriptomes subjected to different strategies for removing redundancy.
Quality MetricTrinity
CD-HIT		Trinity TransPSRockhopper
TransPS
Number of Sequences637338252481
Mean Length479.36743.95687.89
Number with ORF186117581323
Fragments638,046638,046638,046
Mapped602,657 (94%)555,748 (87%)530,315(83%)
TAS0.61220.53030.5405

Share and Cite

MDPI and ACS Style

Caicedo-Montoya, C.; Pinilla, L.; Toro, L.F.; Yepes-García, J.; Ríos-Estepa, R. Comparative Analysis of Strategies for De Novo Transcriptome Assembly in Prokaryotes: Streptomyces clavuligerus as a Case Study. High-Throughput 2019, 8, 20. https://doi.org/10.3390/ht8040020

AMA Style

Caicedo-Montoya C, Pinilla L, Toro LF, Yepes-García J, Ríos-Estepa R. Comparative Analysis of Strategies for De Novo Transcriptome Assembly in Prokaryotes: Streptomyces clavuligerus as a Case Study. High-Throughput. 2019; 8(4):20. https://doi.org/10.3390/ht8040020

Chicago/Turabian Style

Caicedo-Montoya, Carlos, Laura Pinilla, León F. Toro, Jeferyd Yepes-García, and Rigoberto Ríos-Estepa. 2019. "Comparative Analysis of Strategies for De Novo Transcriptome Assembly in Prokaryotes: Streptomyces clavuligerus as a Case Study" High-Throughput 8, no. 4: 20. https://doi.org/10.3390/ht8040020

APA Style

Caicedo-Montoya, C., Pinilla, L., Toro, L. F., Yepes-García, J., & Ríos-Estepa, R. (2019). Comparative Analysis of Strategies for De Novo Transcriptome Assembly in Prokaryotes: Streptomyces clavuligerus as a Case Study. High-Throughput, 8(4), 20. https://doi.org/10.3390/ht8040020

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop