BFAST and miRNA precursor reference

[m_elena_bioinfo]

Dear users,
I'm using BFAST to align miRNA reads from SOLiD ABI to the precursor miRNA reference.

I can't understand one thing.

When I used the reference of miRNA precursor as (for example):
>hsa-mir-548d-1 MI0003668 Homo sapiens miR-548d-1 stem-loop
AAACAAGUUAUAUUAGGUUGGUGCAAAAGUAAUUGUGGUUUUUGCCUGUAAAAGUAAUGG
CAAAAACCACAGUUUCUUUUGCACCAGACUAAUAAAG
>hsa-mir-661 MI0003669 Homo sapiens miR-661 stem-loop
GGAGAGGCUGUGCUGUGGGGCAGGCGCAGGCCUGAGCCCUGGUUUCGGGCUGCCUGGGUC
UCUGGCCUGCGCGUGACUUUGGGGUGGCU
...
...

(extracted by miRBasev13)
the program returns me the error at the localalign step saying me that read and reference don't match.

If I use the same reference, adding to each miRNA sequence 35 N (at the begin and at the end of each one), as, for example:

>hsa-mir-1277
NNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNACCTCCCAAATATATATATATATGTACGTATGTGTATATAAATGTATACGTAGATATATATGTATTTTTGGTGGGTTTNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNN
>hsa-mir-1278
NNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNATTTGCTCATAGATGATATGCATAGTACTCCCAGAACTCATTAAGTTGGTAGTACTGTGCATATCATCTATGAGCGAATAGNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNN
...
...

the program runs and I can terminate my alignment.

So, I don't know how to explain it to myself!

I then compared the number of counts found by BFAST program (about 75000) with the counts found by the RNA_pipeline of corona lite (ABI) (about 22000).
How can I explain this so large difference?
Thank you very much for the help!

Maria Elena

[nilshomer]

Dear users,
I'm using BFAST to align miRNA reads from SOLiD ABI to the precursor miRNA reference.

I can't understand one thing.

When I used the reference of miRNA precursor as (for example):
>hsa-mir-548d-1 MI0003668 Homo sapiens miR-548d-1 stem-loop
AAACAAGUUAUAUUAGGUUGGUGCAAAAGUAAUUGUGGUUUUUGCCUGUAAAAGUAAUGG
CAAAAACCACAGUUUCUUUUGCACCAGACUAAUAAAG
>hsa-mir-661 MI0003669 Homo sapiens miR-661 stem-loop
GGAGAGGCUGUGCUGUGGGGCAGGCGCAGGCCUGAGCCCUGGUUUCGGGCUGCCUGGGUC
UCUGGCCUGCGCGUGACUUUGGGGUGGCU
...
...

(extracted by miRBasev13)
the program returns me the error at the localalign step saying me that read and reference don't match.

If I use the same reference, adding to each miRNA sequence 35 N (at the begin and at the end of each one), as, for example:

>hsa-mir-1277
NNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNACCTCCCAAATATATATATATATGTACGTATGTGTATATAAATGTATACGTAGATATATATGTATTTTTGGTGGGTTTNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNN
>hsa-mir-1278
NNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNATTTGCTCATAGATGATATGCATAGTACTCCCAGAACTCATTAAGTTGGTAGTACTGTGCATATCATCTATGAGCGAATAGNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNN
...
...

the program runs and I can terminate my alignment.

So, I don't know how to explain it to myself!

I then compared the number of counts found by BFAST program (about 75000) with the counts found by the RNA_pipeline of corona lite (ABI) (about 22000).
How can I explain this so large difference?
Thank you very much for the help!

Maria Elena

What version are you using? What post processing options are you using?

[m_elena_bioinfo]

Thanks Dr.Homer,
I'm using bfast-0.6.0d

The options and the parameters, after the fasta2brg and the index step, that I use are:

> bfast match -f hsa_human.fa -r file.fastq -A 1 > file.bmf

> bfast localalign -f hsa_human.fa -m file.bmf -A 1 > file.baf

> bfast postprocess -f hsa_human.fasta -a 3 -O 3 -i file.baf -A 1 > output.sam

With fasta file without NNN, the program crashes at localalign step.

[nilshomer]

Thanks Dr.Homer,
I'm using bfast-0.6.0d

The options and the parameters, after the fasta2brg and the index step, that I use are:

> bfast match -f hsa_human.fa -r file.fastq -A 1 > file.bmf

> bfast localalign -f hsa_human.fa -m file.bmf -A 1 > file.baf

> bfast postprocess -f hsa_human.fasta -a 3 -O 3 -i file.baf -A 1 > output.sam

With fasta file without NNN, the program crashes at localalign step.

Only valid DNA bases are allowed as well as N (so only ACGTN). It looks like you have Us in the reference for the miRNA. Convert those to Ts in your reference.

[m_elena_bioinfo]

Dr. Homer,
in my reference there are not U but it contains only DNA bases. For example:

>hsa-let-7a-2
AGGTTGAGGTAGTAGGTTGTATAGTTTAGAATTACATCAAGGGAGATAACTGTACAGCCTCCTAGCTTTCCT
>hsa-let-7a-3
GGGTGAGGTAGTAGGTTGTATAGTTTGGGGCTCTGCCCTGCTATGGGATAACTATACAATCTACTGTCTTTCCT

So, I don't think that this is the problem!

[nilshomer]

Dr. Homer,
in my reference there are not U but it contains only DNA bases. For example:

>hsa-let-7a-2
AGGTTGAGGTAGTAGGTTGTATAGTTTAGAATTACATCAAGGGAGATAACTGTACAGCCTCCTAGCTTTCCT
>hsa-let-7a-3
GGGTGAGGTAGTAGGTTGTATAGTTTGGGGCTCTGCCCTGCTATGGGATAACTATACAATCTACTGTCTTTCCT

So, I don't think that this is the problem!

Give me your reference and a set of reads an I will test it out myself. Thanks!

Nils

[rdeborja]

Has anyone tried aligning to the mature sequences instead of the entire precursor using BFAST. I've recently jumped on the BFAST bandwagon for genomic data and am trying to find out whether miRNA is feasible. I have two approaches: 1. align to only the mature sequences to identify the known sequences, 2. align to the entire genomic reference and look up and down 100bp of the aligned position for a second hit in the reverse compliment to identify a potential loop structure.