MicroRNA target predictions

→ For more details about our research in microRNA target prediction and for the relevant publications follow this link.

PACCMIT-CDS: A web server for microRNA target prediction

The web server provides a user-friendly interface to two algorithms (see below) for predicting messenger RNA (mRNA) molecules regulated by microRNAs: (1) PACCMIT (Prediction of ACcessible and/or Conserved MIcroRNA Targets), which identifies primarily mRNA transcripts targeted in their 3’ untranslated regions (3’UTRs), and (2) PACCMIT‐CDS, designed to find mRNAs targeted within their coding sequences (CDS). While PACCMIT belongs among the few very accurate algorithms for predicting conserved microRNA targets in the 3’UTRs, the main contribution of the web server is twofold: PACCMIT provides an accurate tool for predicting targets also of weakly conserved or nonconserved microRNAs, whereas PACCMIT‐CDS addresses the lack of similar portals adapted specifically for targets in CDS.

The web server asks the user for microRNAs and mRNAs to be analyzed, accesses the precomputed p‐values for all microRNA-­mRNA pairs from a database for all mRNAs and microRNAs in a given species, ranks the predicted microRNA-­mRNA pairs, evaluates their significance according to the false discovery rate, and finally displays the predictions in a tabular form. The results are also available for download in several standard formats. For a more detailed description see:

 

M. Šulc, R. M. Marín, H. S. Robins, and J. Vaníček, Nucleic Acids Research 43, W474-­W479 (2015)

PACCMIT-CDS algorithm (November 2012)

This algorithm predicts microRNA targets in coding sequences and was published in

 

R. Marín, M. Šulc, J. Vaníček, RNA 19, 467-474 (2013).

For the source code, predictions, and details about this algorithm, follow this link.

PACCMIT predictions (September 2012)

If you use this algorithm, please cite R. Marín et al. RNA 18, 1760-1770 (2012).
For details, see the brief description of the algorithms at the bottom of the page.

Dataset Organism Method* Description
1 Human PACCMIT Accessibility (restricted location) These predictions have been obtained by restricting the location of the nucleation region, i.e., only seed matches with accessible 4-mers pairing miRNA positions 2-5 are considered. This was shown to increase the precision of the algorithm with respect to the case in which any accessible 4-mer within the seed match was enough to label the site as accessible. An optimzed Pcutoff of 0.2 was used.
2 Human PACCMIT Access + Cons (restricted location)

* Predictions are for all miRNAs in miRbase v18.

PACCMIT predictions (February 2012)

If you use this algorithm, please cite R. Marín and J. Vaníček, PLoS ONE 7, e32208 (2012).
For details, see the brief description of the algorithms at the bottom of the page.

Dataset Organism Method* Description
1 Human PACCMIT No Filter These are the high-sensitivity predictions. Neither of the filters is used. All seed matches are considered.
2 Human PACCMIT Accessibility (unrestricted location) These are considered to be the high-confidence predictions for weakly conserved miRNAs. A cutoff of Pcutoff = 0.2 was used (i.e., only seed matches containing 4-mers accessible with probability of at least 0.2 were considered).
3 Human PACCMIT Conservation These are considered to be the high-confidence predictions for highly conserved miRNAs. Only seed matches conserved among human, chimp, rhesus, and mouse were considered.
4 Human PACCMIT Access + Cons (unrestricted location) These are considered to be the high-confidence predictions for highly conserved miRNAs. Only seed matches meeting both “Accessibility” and “Conservation” requirements were considered.

* Predictions are for all miRNAs in miRbase v18.

PACMIT predictions (August 2010)

If you use this algorithm, please cite R. Marin and J. Vanicek, Nucleic Acids Research 39 19-29 (2011).
For details, see the brief description of the algorithms at the bottom of the page.

Dataset
Organism Method* Description
1 Human PACMIT-0.0 These are the high-sensitivity predictions. A cutoff of  Pcutoff = 0.0 was used (i.e., the filter was not used). All seed matches were considered.
2 D. melanogaster
3 Human PACMIT-0.2 These are considered to be high-confidence predictions. A cutoff of Pcutoff = 0.2 was used (i.e., only seed matches containing 4-mers accessible with probability of at least 0.2 were considered).
4 D. melanogaster

* Predictions are for all miRNAs in miRbase v13.

Brief description of the algorithms

Steps to predict microRNA targets with PACCMIT:

1) Look for “seed matches”, i.e., the sites in the target sequence which are perfectly complementary to the microRNA seed. The “seed” consists of the nucleotides between (and including) positions i and j in the microRNA. All results reported here use the standard choice i = 2 and j = 8.

2) Filter the seed matches according to one or the two following criteria.
  • Accessibility (unrestricted location): for all 4-mers in the seed match, compute the probability Pu that the 4-mer is accessible (or “unpaired”), i.e., contained within a single-stranded region of RNA. Select only the “partially accessible”  seed matches, i.e., those that contain at least one 4-mer with Pu >= Pcutoff.
  • Accessibility (restricted location): seed matches are considered as “partially accessible” only in cases where the 4-mer opposite to positions 2 to 5 of the miRNA has Pu >= Pcutoff.
  • Conservation: Select only the seed matches that are conserved in the human, chimp, rhesus, and mouse 3’UTRs.

3) For each target sequence compute the number cfilter of seed matches that meet the filter requirements.

4) Compute the probability PSH to observe at least cfilter seed matches in a random background sequence with the same dinucleotide composition as the real sequence. PSH measures the over-representation of the seed matches in a given target sequence, therefore lower PSH values (higher over-representation) indicate higher chances of biological functionality. The predictions provided above are always ranked by over-representation.

NOTE: The PACCMIT algorithm (PLoS ONE, 7, e32208, 2012) makes reference to the method in which both accessibility and conservation filters are implemented. The PACMIT algorithm (Nucl. Acids Res., 39, p 19-29, 2011) refers to the method in which only the accessibility filter is implemented.