Use blastn to predict miRNA
3
0
Entering edit mode
7.5 years ago
MeiNB ▴ 10

Hi,

I am using blastn against mature miRBase, to obtain the conserved miRNA. Then, using the reads that don't align in the previous step I will use blasn against Rfam, to remove all non-coding RNA that aren't miRNA. The final step, is align the left reads against the genome.

What you think about this approach?

miRNA blastn new miRNA conserved miRNA • 2.2k views
ADD COMMENT
0
Entering edit mode

What is your starting input? (small RNA-seq fastq?) and what is exactly your goal? (find new miRNAs which are not highly conserved between your specie of interest and a database?)

ADD REPLY
0
Entering edit mode

Hello, please post on THIS comment tread instead of adding an answer as it is difficult to follow, just hit ADD REPLY :).

Anyhow, it can be done using Blast, but I would use Bowtie due to efficiency and it being built for small reads. Also, be wary of miRBase redundancy, as some miRNAs being shorter than other for any reason (for example, IsoMiRs) will be mapped as something else and you will end up with more miRNAs than you should actually have.

Also, if you are looking for the right tool to do it, THIS PAPER should have all the answers you are looking for. Cheers and good luck! :D

ADD REPLY
1
Entering edit mode

Hi, you are right ... sorry for my distraction. From now on I will do that. :)

My first approach was use bowtie, but my results were worse than blastn results. Initially, the average of alignment was 10%, then I modified the parameters of bowtie to increase the aligned reads (I took into account the T/U, I preprocessed the reads and checked that the adapters were removed) and I get almost 30%. When with blastn I obtain 80%.

I used: bowtie -n1 -l8 -a (my bowtie's runs with the parameters --best --strata didn't finish ... :( ) and bowtie -v 1 -k1. I tried more but with these two I get more % of alignment. Maybe I am too stringent?

I didn't mention but I filter the miRBase for plants. I am aware of miRBase redundancy and one of my fear is that, get more miRNAs that was supposed.

Thank you for the encouragement. I need!

ADD REPLY
0
Entering edit mode
7.5 years ago
MeiNB ▴ 10

Sorry, I should have given more information.

My data are small-RNA sequenced using the Illumina platform. I preprocessed the reads (output format was fq) and then align with blastn.

Yes, one of my goals is to identify new miRNAs and conserved miRNA. I have on variety of a specie, and I want to find possible new miRNAs between them.

ADD COMMENT
1
Entering edit mode

Since you have HTS data you should have a look to these tools to find new miRNAs: https://omictools.com/ncrna-identification-category and probably this publication: https://www.ncbi.nlm.nih.gov/pubmed/23334922

if you still want to try to blast your data I think your method is a good starting point. For your last step you will have to use an aligner such as bowtie to map them onto the genome.

ADD REPLY
0
Entering edit mode
7.5 years ago
MeiNB ▴ 10

Thank you for your help :)

Some of these tools, like mirdeep2, I already tested but the results are worse than the results from blast. I will see the others software and then test the ones that I think that are more suitable for my data.

So, I will follow your advice and map with bowtie (I read that bowtie1 is better than bowtie2 to map short-reads), but I also want try another tool, to compare the results, in your opinion BWA is a good choice?

ADD COMMENT
1
Entering edit mode

For small rna use bowtie1. This usually gives me good results. Sometimes simplier approaches are the best. If the results make sense and can be validated after your are in buisness.

ADD REPLY
0
Entering edit mode
7.5 years ago
MeiNB ▴ 10

I will read the manual and search which are the best parameters to run bowtie1 with short-reads.

Once again, thank you for your help!

ADD COMMENT

Login before adding your answer.

Traffic: 1496 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6