identify new miRNA

Question

[nasromer2191989]

hi Is there possibility to identify new miRNA associated with certain diseases by building model for the once that already Known? thanks

[WouterDeCoster]

Based on which type of data? Your question lacks information. It's unclear what you have and what you need.

[nasromer2191989]

based on the previous known miRNA that linked to the disease.I have there name and sequence. What I would like to do is to predict a new miRNA that have similar function to the known miRNA and check if that new miRNA is in the cell line be informed that all miRNA extracted from different database and the sequence collected from miRBase.

Answer 1

[nasromer2191989]

based on the previous known miRNA that linked to the disease.I have there name and sequence. What I would like to do is to predict a new miRNA that have similar function to the known miRNA and check if that new miRNA is in the cell line be informed that all miRNA extracted from different database and the sequence collected from miRBase.

[Prasad]

You can align all reads mirbase to filter out known, then predict novel by aligning to genome (using mireap or mirdeep2 etc). do the target prediction for the novel ones, then you can link the disease based on targets

[Farbod]

Dear Prasad, Hi

About the "align all reads [to] mirbase [database]", how it could be done ?

Is it via blastN ?

if yes what e-value must to be used ?

[Prasad]

You could use bowtie1 end2end with 1 mismatch or SHRiMP which has specific miRNA mode. Even blast could work.

[Farbod]

Dear Prsasad, Hi and thank you for your help

Is there any script (with more details) you could offer or any paper that describe this script, because the bowtie accept many parameters and options.

I have found this post and this one but it seems that it is about mapping the miRNA to genome, I think we must use script for aligning our miRNA fastq to mirbase data, yes?

And I guess we must convert our fastq(s) to fasta and then use bowtie but by which script and options and database (Mature mirbase or hairpin mirbase) ?

please have a look on these scripts from here:

Local bowtie2 alignment of miRNA data

(I generated the reference with bowtie-build directly from the hairpin FASTA file downloaded from miRBase.)

Bowtie2-build hairpin_mms.fa hairpin_mms.fa

Bowtie2-build mature_mms.fa mature_mms.fa

Bowtie2 –local –N 1 –L 16 –x hairpin_mms.fa –U fastq/xxx_microRNA(_adaprm).fastq  -S xxxx.sam

[Prasad]

have you tried mirdeep2.

[Farbod]

yes but unfortunately I have no reference genome (even no close species), and it needs to map the reads to the genome !

I just have 6 set of miRNA fastqs (3 for male and 3 for female as biological replication) and 6 set of RNA-seq for the same individuals (and a de novo transcriptome assembly).

So, I first intend to collect the previously known miRNAs (using blastn or bowtie, if possible) and then focus of the target of the novels and perform some DEmirna analysis.

~ Best

[Prasad]

if you done have the genome, how will you predict novel miRNA. If you intend to do it against transcriptome data mirdeep could be an option.

[Prasad]

in my opinion, other option would be- trim the data with srnaworkbench for trimming (as it generates fasta with read count) or any other. Map reads to miRBase sequences, take only those with alignment length >=18 with least mismatches (non or 1 would be preferred). Then DE can be performed using DESeq2 or edgeR.

[Farbod]

Very nice, thank you for all your supports.

could you please kindly suggest some script for "Map reads to miRBase sequences, take only those with alignment length >=18 with least mismatches" ? If possible for you, I want something like this : bowtie bla..bla..bla all-mature-mirbase.fa –U my--miRNA-reads.fastq . . . and so on

as I have 6 sets of data, should I merge all males (3 fastqs) and all females (3 fastqs) with each other and then use the mapping script one time for males and one for females (two times) ? or run it 6 times (one for each) ? or concatenate all the 6 fastqs into one file and then Trimming and then mapping ?

[Prasad]

run individually as they are biological replicates

[nasromer2191989]

sorry I have question if use blastn and using the list of known miRNA as query to EST & GSS database the result file what would be? if the similarity is 100% is that mean the miRNA I send it as query is already identified and I try that and I couldn’t turn the subject sequence to miRNA while I am looking for miRNA

is there any method that I can follow to predict new miRNA for speicifc disease to clear out my way in this project? please help thanks all

[Prasad]

go through this, hope this answers

[nasromer2191989]

As I am new in this field could you explain more please. How to do align all reads of mirbase?

[Farbod]

There are some help here, but the e-value is still a problem. some use it as 10 but I can not understand what does it mean? is it -evalue 1e-10 or -evalue 10 or . . . ?