wapRNA: a web-based application for the processing of RNA sequences

Abstract

1 INTRODUCTION

RNA-Seq is a recently developed revolutionary approach for transcriptome profiling and miRNA analysis using next generation sequencing technologies. It can allow us to study RNA in further depth and more accurate measurement. Studies using this method have already altered our view of transcriptome and miRNA. Recently, a lot of methods have been developed to process the different aspects of RNA-seq data, such as short reads mapping (Li and Durbin, 2009), gene annotation and differential gene-expression analysis (Wang et al., 2010). Meanwhile, several web-services for RNA and microRNA analysis have also been developed, such as rQuant.web (Bohnert and Ratsch, 2010), MAGIA (Sales et al., 2010), miRCat (Moxon et al., 2008), miRAnalyzer (Hackenberg et al., 2009), mirTools (Zhu et al., 2010) and DSAP (Huang et al., 2010). However, some of these web services need prior data processing, such as mapping raw sequences, normalizing expression data, or they do not provide local installation package.

In order to provide a service for the increasing amount of experimental studies using RNA-seq methods, here, we present the wapRNA, a novel web service that allows us to do mRNA and miRNA expression analysis simultaneously. wapRNA complements other RNA analysis tools reported by Goncalves et al. (2011). Moreover, we provide the executable packages for installation on user's local server. This will facilitate studies for RNA using high-throughput approaches, especially for biologists who have limited computing sources or are short of bioinformatics processing experience.

2 METHODS

2.1 The analysis pipeline

The wapRNA web service simplifies the analysis of mRNA-seq/miRNA-seq data by executing a computational pipeline on a set of data uploaded by user. To accommodate different sequencing methods (SOLiD or Solexa) and study targets (RNA or miRNA); we built four modules for SOLiD mRNA, Solexa mRNA, SOLiD miRNA and Solexa miRNA data analysis, respectively. Ten animal species whose genome sequences are available are provided in our server (The species will be added according to user's requirements in future), and the gene structure information is based on ENSEMBL (http://asia.ensembl.org/index.html) annotation. For mRNA data analysis, the process starts from selecting sequencing platform, uploading raw data and specifying the parameters like species and mapping mismatches allowance. When submitted the above request, wapRNA will launch the processes by preparing reference sequences and filtering raw data according to user's request or default parameters. Then, those cleaned reads will be mapped to corresponding reference sequences automatically using Corona_Lite package for SOLiD (http://solidsoftwaretools.com/gf/project/corona/) and BWA package (Li and Durbin, 2009) for Solexa, respectively. The expression levels (RPKM, Reads Per Kilobase of exon model per Million mapped reads) for each gene will be calculated using uniquely mapped reads by in-house built Perl module. Moreover, the downstream functional classification will be launched through the integrated localization of Gene Ontology (Harris et al., 2004) and KEGG pathway database (Kanehisa and Goto, 2000). When the job is finished, an email alerts will be automatically sent if the email address is provided, and all results can be found and downloaded at the results page, and also some concise pictures and statistical tables are provided for users.

While for miRNA data analysis, users also should specify the requisite parameters listed in webpage and submit the job. We adopted two different methods for SOLiD and Solexa data analysis. For Solexa platform, we first remove adaptors sequences, cluster those identical reads into unique tags, map tags to assigned genome sequences and use those perfect hits for the following analysis. Then we filter those non-miRNA (including snRNA, tRNA, rRNA, mRNA, etc.) tags and annotate conserved miRNA by comparing tags to miRBase (Griffiths-Jones, 2006; Griffiths-Jones et al., 2008) database. Finally, we predicate new miRNAs using miRdeep software (Friedlander et al., 2008). While for SOLiD platform, due to the color-spaced sequences, we integrate the RNA2MAP package (http://solidsoftwaretools.com/gf/project/rna2map/) for filtering the non-miRNA sequences, annotating the known miRNAs by aligning to the miRNA precursor sequences from miRBase, and mapping to the genome sequence for predicting novel miRNAs. The miRNA targets are predicated and combined by using RNAhybrid (Kruger and Rehmsmeier, 2006) and miRanda (Betel et al., 2010) software.

Moreover, we developed an independent module for detecting differentially expressed genes (DEG), functional classification (based on GO and KEGG database), and miRNA target predication.

2.2 Case study

To demonstrate the functionality of wapRNA, we used it to analyze SOLiD mRNA data from SRA009022 (Cui et al., 2010) and miRNA data (Cai et al., 2010) produced in our lab. We also tested our Solexa pipeline using mRNA data from SRR037165 (Wu et al., 2010) and Solexa miRNA data from SRR026761 (Morin et al., 2008). All analyzed results were showed in demo results part, and all these jobs were run on a Red Hat Enterprise Linux Cluster with 12 16 GB-RAM compute nodes and a total of 96 CPU cores.

For the SOLiD mRNA pipeline, we got 13.6% multiple mapped reads and 32.8% unique mapped reads, which are relatively higher than Cui et al.'s (2010) original results. The difference is caused by parameter setting in mapping procedure. For the miRNA pipeline, we identified 397 (341 unique) known miRNA and 138 (111) novel miRNA in only one sample (hESC), while 315 known miRNA and 100 novel miRNA in hESC samples by Morin et al.'s (2008) results, about 82% of those known miRNA and 87% of those novel miRNAs are overlapped between the two methods. The other two tests also showed very similar results with original results.

3 IMPLEMENTATION

wapRNA consists of front- and back-end programs. The front-end program, constructed based on Java Server Page (JSP) 2.0 technology, is in charge of tasks submission and results display. We used the Struts (http://struts.apache.org), Spring (http://www.springsource.org) and Hibernate (http://www.hibernate.org/) framework to enhance the flexibility and extendibility, and Mysql5.0 as the DBMS to store the interim and final results. The back-end program consists of series of Perl and Shell scripts, it is mainly in charge of the data processing.

The executable packages for four pipelines designed and implemented using Java program, which can be download from our web site. All the packages can be run on 64-bit Linux system. Before running the packages, some essential tools declared in the download page are needed to be pre-installed in the system.

4 DISCUSSION

The wapRNA web service provides users to analyze next-generation sequencing mRNA and miRNA data directly and simply by an easy-to-use browser user interface. The service includes major processes for the next-generation mRNA or miRNA data analysis, including preprocessing raw sequenced reads, mapping tags to reference sequences, gene expression annotation, and other downstream functional analysis such as detecting differentially expressed genes, Gene Ontology and KEGG pathway analysis for RNA, novel miRNA predication and miRNA target prediction. Moreover, we also provide executable packages for users to build their pipeline locally. However, some steps are still improvable, for example, we ignored those multiple mapping reads which may be very important for duplicated genes or members of gene family. And also, there are a few additional features we have not involved, such as new transcript identification, alternative splicing, these features will be integrated into the wapRNA in the near future.

ACKNOWLEDGEMENTS

We would like to thank all the members of Genome and Bioinformatics platform in BIG for their assistance during the pipeline development, and thank Peng Cui, Xiaomin Yu for the previous work on the RNA and miRNA research, and also thank Kesheng Liu for the solid data support.

Funding: Natural Science Foundation of China (30900831 to S.S.); Ministry of Science and Technology of the People's Republic of China (BSDN2009-15 to W.Z.); Knowledge Innovation Program of the Chinese Academy of Sciences (KSCX2-EW-R-01-04 to S.H.).

Conflict of Interest: none declared.

REFERENCES

Author notes