Linking Mirnas With Disease Or Keywords

Question

[Ryan D]

We have a list of about 30 miRNAs with accession number and name like so:

MIMAT0003313    miR-643
MIMAT0000446    miR-127-3p
MIMAT0002814    miR-432
MIMAT0000423    miR-125b
MIMAT0003214    miR-551a
MIMAT0004694    miR-342-5p
MIMAT0003284    miR-616*

I want to try to connect those miRNAs to diseases or key words--such as "circulating", "plasma", or "cancer". Programs like chilibot can do this well for genes since gene/protein names and aliases are well-mapped. I wonder if it is possible to do this for miRNAs, though. miRNA id's, family names, or aliases do not seem to have such a tool available as I can tell. If something exists which can mine Pubmed or other databases to look for matches between miRNAs in this list and certain keywords/diseases, that would be great. If there is another handy way you can think to do this, I'd love to hear suggestions.

Answer 1

[Anthony Mathelier]

I think you will be interested in having a look at PhenomiR (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2847718/).

[Ryan D]

The paper makes it sound as though at least for the disease-miRNA link, that this is the proper tool, though perhaps not for keywords. It seems that the link to PhenomiR http://mips.helmholtz-muenchen.de/phenomir is down at the moment. Hopefully it is not another stale bioinformatics resource. Thanks.

[Anthony Mathelier]

If down you should email the authors but I guess it should fullfill your needs at least for the disease part. And you can also ask the authors if they are aware of other tools.

Answer 2

[Neilfws]

You can use EUtils ESearch to search the NCBI nucleotide database for miRNAs, then link to PubMed using ELink.

Here's a quick example using the BioRuby implementation of EUtils. Search this site too (e.g. for "elink"); the topic has been covered before.

To search the Nucleotide database:

#!/usr/bin/ruby
require "rubygems" # ruby 1.8.7
require "bio"
require "open-uri"
require "crack" # for XML parsing

Bio::NCBI.default_email = "[e-mail]"
ncbi = Bio::NCBI::REST.new
ncbi.esearch("miR-643", {"db" => "nucleotide", "retmax" => 200})

# => ["301172193", "270132492"]

To link nucleotide GI with PubMed UIDs:

# use ID 301172193 as example
base = "http://eutils.ncbi.nlm.nih.gov/entrez/eutils/"
suff = "elink.fcgi?dbfrom=nucleotide&db=pubmed&id=301172193"
xml  = open("#{base + suff}").read
xml  = Crack::XML.parse(xml)
# ugly code to get the PubMed IDs
ids = xml['eLinkResult']['LinkSet']['LinkSetDb'].map {|ls| ls['Link'].map {|l| l['Id']} }
# => [["18760970", "16381832"], ["18760970", "16381832"]]
# more ugly code for unique IDs
ids = ids.flatten.uniq
=> ["18760970", "16381832"]
# do something with IDs e.g. EFetch

[Ryan D]

Thank you. This looks like an inventive solution. I guess it depends on PubMed entries having good identifiers for miRNAs--something I'll need to look into.

Answer 3

[Ryan D]

My colleague, Jingchun Sun, had some useful links toward this end including: HMDD The human microRNA disease database (HMDD) and miR2Disease base.

The former of them claims to be more up to date than the latter. I will update once I know more.

[Khader Shameer]

+1 for HMDD; they also provide PMID in one of the downloadable files.

[Ryan D]

That's news I can use, Khader. Thanks.