Web demos mirDNN

Resources

Source code: (1) | Data: {1} {2} | Related publications: [submitted]

mirDNN is a deep-learning based method for pre-miRNA prediction based on the RNA sequence. This demo provides pre-trained models, with known pre-miRNAs from animals or plants, to predict candidate sequences to be novel pre-miRNAs.

If a fasta file is provided, a score for each sequence within the file is returned. Sample inputs can be downloaded for animals and plants.

If a single sequence is provided, the output is the prediction score and the importance given by mirDNN to each nucleotide in the sequence for the classification decision.

  • Sample sequence for animals model:
    ugggaugagguaguagguuguauaguuuuagggucacacccaccacugggagauaacuauacaaucuacugucuuuccua
  • Sample sequence for plants model:
    guguugacagaagauagagagcacagaugaugagauacaauucggagcauguucuuugcaucuuacuccuuugugcucucuagccuucugucaucacc

Contact: Cristian Yones