When we are taking about non-coding RNA (ncRNA) we are primarily meaning long non-coding RNAs (lncRNAs) and microRNAs (miRNAs). Here we describe some important points in our workflow analysing microRNA-Seq (miRNA-Seq) experiments.
We are analyzing RNA-Seq data from all mayor next-generation sequencers from Illumina or Ion-Torrent. We can start from files in FASTA, FASTQ, unaligned BAM files or SRA format.
Please scroll down for more information about the single steps of our RNA-Seq workflow. Please contact us here, in case you have any question about our service.
Quality control and read preparation
The quality control of raw data (FastQ files) of miRNA-Seq experiments is analogue to what we describe in our standard RNA-Seq section of our portfolio. However subsequent read preparation differs strikingly. Since mature micro RNA are short RNA fragments of just about 22 nucleotides, adapters had to be introduced during the NGS assay. In order to be able to align the sequence tags properly, those adapter sequences had to be trimmed away completely without interfering with the actual miRNA sequence.
Alignment, normalization and quantification
Alignment of miRNA-Seq data follows different rules then of normal RNA-Seq data. Most strikingly we do not need to use a splice-aware aligner like Star or Tophat. We do have very good experience with using BWA for most of our miRNA-Seq experiments, however it might be necessary to check other options at this stage. Alignment can be done using the whole genome as reference or directly the set of mature or immature miRNA sequences. We prefer the first option, primarily because we can later check for novel miRNAs, which are not listed in the micro RNA databases. Quantification is then usually done using mature miRNA sequences from trusted miRNA databases. Normalization of micro RNA results can be trickier than normalising mRNA data. Highly specific experimental designs can alter the expression of a subset of miRNAs so drastically, that some normalisation procedures are just not valid. We know how to deal with this issue. And we have a list of quality matrices in place to validate the result of the normalisation procedure.
If asked for, we additionally perform detection of possible novel miRNAs which we add to the results of the standard analysis.
We do take the statistical detection of differentially expressed non-coding RNAs very serious. We perform a meta-analysis of the popular statistical algorithms for NGS expression data and carefully select the best performing one. Please have a look at our RNA-Seq section in the portfolio for one example.
mRNA target prediction and integration with gene expression data
At this step we already have prepared dozens of high-quality figures for you, helping you to interpret the outcome of your experiment. In a next step we associate your significant regulated miRNAs with their dedicated mRNA targets. We will discuss with you the stringency of the selection of mRNA target. Selection is based on the probability being a real target (for example confirmed by experimental evidence only) and this should be adjusted to the aim of the subsequent exploration. On standard investigation would be the detection of over represented target genes in distinct Pathways or Gene Ontolgoy categories.
In case you have also corresponding mRNA results at hand, we merge the two data sets. This enables us to make for example correlation analysis of miRNA:mRNA target sets.