Determination of target genes of DE miRNAs and interpretation of the sequencs?

Question

[top5]

Hello, I recently worked on predicting target sites of DE miRNAs by using miranda v3.3a but i get just ref sequences and alignment score etc. Then i used 'blastn' to find which genes are aligned for the ref sequences that i found in miRanda. But do you think i did right way? I do not want to do something wrong if anyone knows it could you please answer?

Answer 1

[Kevin Blighe]

Hi, a simpler / quicker way is likely via miRNAtap. In this example, I search for targets of [hsa]-miR-151-5p and I aggregate the findings across PicTar, DIANA, TargetScan, and miRanda. This program will rank the target genes.

require(miRNAtap)
require(miRNAtap.db)

searchTerm <- 'miR-151-5p'

mirTargetPredictions <- getPredictedTargets(
  searchTerm,
  sources = c('pictar','diana','targetscan','miranda'),
  species = 'hsa',
  method = 'geom',
  min_src = 3)

head(mirTargetPredictions)
       source_1 rank_product rank_final
27076         1            1          1
10235         2            2          2
25769         3            3          3
57572         4            4          4
171023        5            5          5
728642        6            6          6

Those rownames are Entrez IDs. We can convert these to HGNC symbols and Ensembl gene IDs:

require(biomaRt)
mart <- useMart("ENSEMBL_MART_ENSEMBL")
mart <- useDataset("hsapiens_gene_ensembl", mart)
annot <- getBM(
  mart = mart,
  attributes = c('entrezgene_id', 'hgnc_symbol', 'ensembl_gene_id'),
  uniqueRows = TRUE)

mirTargetPredictions <- data.frame(
  mirTargetPredictions,
  annot[match(rownames(mirTargetPredictions), annot$entrezgene_id),])

head(mirTargetPredictions)
       source_1 rank_product rank_final entrezgene_id hgnc_symbol
27076         1            1          1         27076       LYPD3
10235         2            2          2         10235     RASGRP2
25769         3            3          3         25769     SLC24A2
57572         4            4          4         57572       DOCK6
171023        5            5          5        171023       ASXL1
728642        6            6          6        728642      CDK11A
       ensembl_gene_id
27076  ENSG00000124466
10235  ENSG00000068831
25769  ENSG00000155886
57572  ENSG00000130158
171023 ENSG00000171456
728642 ENSG00000008128

Please explore the various parameters that are being used with these functions.

Kevin

[top5]

Thank you so much your answer but my species is rabbit that is why i used miranda to scan its reference genome, the package that you suggest does not include this species...

[Kevin Blighe]

No problem, but, unfortunately, Oryctolagus cuniculus was not mentioned in your question.

Would it not also be a good idea to align the sequences to the annotation that already exists? - https://www.ensembl.org/Oryctolagus_cuniculus/Info/Index (see 'gene annotation' section).

[top5]

Yes i could do that but miRanda gives target sequences of genes. My main question is, after I got the target sequences, I used blastn to find out which genes these sequences belong to, is this the correct way? Or do you know another packages to do that? Because i have lack of information about it obviously...

[Kevin Blighe]

miRanda returns the predicted target mRNA sequence of your input micro-RNAs, right?

You can use BLAST for these, but I see no reason why you cannot also use the data that is already available for this species, e.g., aligning the predicted target sequences against the gene FASTA data available at: https://www.ensembl.org/Oryctolagus_cuniculus/Info/Index

There is no major issue with your approach, provided that you understand how to filter the BLAST output for the best alignments.

[top5]

Now i understand what do you mean, yes you are right i will try it now !! Thank you so much. I didn't think in the first place.