Welcome to miREV

Extracellular vesicles
Extracellular vesicles (EVs) such as Exosomes are small membrane-enclosed vesicles that are released in all body fluids by a variety of cells. Due to their aberrant expression under different physio- and pathophysiological conditions and the important role they play in intercellular communication, miRNAs have emerged as a promising source of biomarkers. In addition to their specific protein and DNA cargo, EVs have been shown to be specifically enriched in miRNAs . EVs from biofluids in general and blood in particular comprise a multitude of morphologically different subclasses and originate from many different cell types. Therefore, experimental conditions such as EV isolation methods have a major impact on downstream RNA profiles and comparability of results from different studies even from the same tissue is low.

miREV database
miREV contains data from publicly accessible sources as well as user submitted, unpublished data. Focusing on blood derived EVs, miREV v 0.8.2 includes 9 different pathologies and 3 different isolation methods from both serum and plasma so far, which will be expanded upon with future data additions. Strict emphasis was put on data sets performing validation of EV isolates by complementary methods such as Western blot or nano particle tracking analysis. To ensure comparability and to exclude poor quality data, sequencing data was processed using a standardized pipeline. In total 654 samples from 13 studies were incorporated of which 428 samples remained after the strict filtering procedure. To depict a multitude of analysis setups, count lists were normalized by 6 commonly used normalization methods for abundance analyses and potential, stably expressed reference transcripts were evaluated by 3 different, established algorithms. Raw count tables as well as the associated metadata for the selected sample set can be downloaded directly from this website. The code for the analysis pipeline including further details is available on GitHub.

General information:
Please be patient if selections aren’t visible immediately since it takes the server a short time to load the data. Experimental variables can be selected by clicking directly in blank fields or chosen from the drop-down menu beneath each category. Selections can be undone by highlighting them first and then using the delete key. In the following, the individual analysis steps (tabs on the right) and their purposes are explained.

Experimental setup:
First the user needs to create a sample set based on his choices for EV-isolation method, biofluid of origin and related disease. To avoid selections of combinations for which no data is incorporated in miREV, yet, users have to state their preferences step-by-step, starting with isolation methods, then biofluids and ending with a disease state. Matching samples will be fetched from the miREV database and form the individual basis for the following analysis steps (PCA and reference miRNA detection). Sample set composition can be reviewed in a tabulary overview after selection of all variables.

PCA & Raw data
The relationship between individual experimental variables and the major sources of variance within the sample set can be investigated on this tab by principal component analysis. Samples with similar expression profiles will cluster in close proximity and can be colored according to the chosen variables of the sample set to highlight influences.
Furthermore, raw read counts of the chosen sample set as well as corresponding meta data can be downloaded from this tab as well. This allows users to run further downstream analysis such as differential gene expression without the need for a time-consuming gathering and bioinformatically demanding alignment and annotation of small RNA-seq data.

Stably expressed miRNAs
Stability of reference transcripts is heavily influenced by normalization and stability calculation strategies, especially when combining data from different studies. Since there is no single best normalization method, miREV offers the opportunity to normalize by multiple different algorithms and more importantly offers consensus summaries to pick miRNAs displaying high potential for references in multiple scenarios. Results are presented in 3 different tabs:
Occurrence of transcripts will show how often potential reference transcripts are listed in the result lists, a combination of one normalization method and one stability measure algorithm). For example, if the user chooses two normalization method and three stability measure algorithms, six result lists in total are taken into account.
Compare stability algorithms shows the overlap between result lists for all selected stability measure algorithms in an easy to interpret Venn diagram. The user can see which transcripts are found commonly in all algorithms or uniquely in some.
Compare normalization mehtods shows, similar to the previous tab the overlap analysis for all selected normalization methods.

miREV is interactive.
All plots and tables are changed when the user changes experimental or calculation variables.

Please select a group from the dropdown menu that you would like to compare using the PCA plot.
If an error messange appears please check if your sample set has more than 10 samples.