an error from running with mouse databases

[Hoon Kim]

Dear Eduardo,

After downloading 'ftp://mirbase.org/pub/mirbase/CURRENT/genomes/mmu.gff3' and 'ftp://ftp.ccb.jhu.edu/pub/data/bowtie_indexes/mm9.ebwt.zip', I tried to generate 'ReadCount results'. But, the output is different from the one I had from your Example data: S177_ATCGTT S178_GACGTT S179_AGAGTT S180_GTTCTT S83_ATGTTT S84_GAGTTT S85_AGCTTT S86_GTATTT S89_ACATTT S90_GGTGTT S91_AATGTT S92_GCGGTT 5 5 23 6 41 12 71 69 7 44 15 61

It looks like my mouse mmu.gff3 may not be compatible with the mirrma program. I wonder if you can provide your comments on how to solve this program.

thank you in advance,

Hoon

[Eduardo Andres Leon]

Dear Hoon, All kind of GFF are compatible with miARma as GFTT files, are standard annotation files. I wonder which parameters did you use. For this mmu.gff3, you will need:

 [ReadCount]
 database=path_to_mmu.gff3
 ;GFF attribute to be used as feature ID (default: gene_id) for featureCounts analysis
 seqid=Name
 ;Feature type (3rd column in GFF file) to be used, all features of other type are ignored (default:exon) for featureCounts analysis
 featuretype=miRNA

greetings

[Hoon Kim]

Thank you very much for your comments. Now it works.

By the way, I think your program can also measure # of reads on other small RNAs(<100 bps) if the corresponding gtf is provided. Is it correct?

thank you,

Hoon

[Hoon Kim]

Dear Eduardo,

I have another question. After creating an annotation GTF consisting of all small mouse transcripts (<=100 bps), I ran miarma.

a part of 'summary_results.xls' is:

Alignment [/scratch/bcb/hkim6/SH-mouse-miRNA/work/merged-fastq/miARmaSeq/Known_miRNAs/results_mmu.NCBIM37.67_100bp.gtf//Bowtie1_results/]
Filename    Processed Reads Aligned reads   Failed to align
S177_ATCGTT.fastq   1189389 893362 (75.11%) 296027 (24.89%)
S178_GACGTT.fastq   768298  580088 (75.50%) 188210 (24.50%)
S179_AGAGTT.fastq   1183877 866511 (73.19%) 317366 (26.81%)
S180_GTTCTT.fastq   1378839 1039826 (75.41%)    339013 (24.59%)
S83_ATGTTT.fastq    2264620 1633403 (72.13%)    631217 (27.87%)
S84_GAGTTT.fastq    1422850 1108686 (77.92%)    314164 (22.08%)
S85_AGCTTT.fastq    2364015 1656693 (70.08%)    707322 (29.92%)
S86_GTATTT.fastq    2570765 1656320 (64.43%)    914445 (35.57%)
S89_ACATTT.fastq    1338242 1084348 (81.03%)    253894 (18.97%)
S90_GGTGTT.fastq    2634816 1691761 (64.21%)    943055 (35.79%)
S91_AATGTT.fastq    1121329 828391 (73.88%) 292938 (26.12%)
S92_GCGGTT.fastq    2574415 1778573 (69.09%)    795842 (30.91%)

ReadCount [/scratch/bcb/hkim6/SH-mouse-miRNA/work/merged-fastq/miARmaSeq/Known_miRNAs/results_mmu.NCBIM37.67_100bp.gtf//Readcount_results/]
Filename    Processed Reads Assigned reads  Strand  Number of identified entities
S177_ATCGTT_nat_bw1 1189389 57491 (4.8%)    no  399
S178_GACGTT_nat_bw1 768298  21267 (2.8%)    no  347
S179_AGAGTT_nat_bw1 1183877 62632 (5.3%)    no  415
S180_GTTCTT_nat_bw1 1378839 43788 (3.2%)    no  411
S83_ATGTTT_nat_bw1  2264620 135825 (6.0%)   no  453
S84_GAGTTT_nat_bw1  1422850 43822 (3.1%)    no  387
S85_AGCTTT_nat_bw1  2364015 159310 (6.7%)   no  440
S86_GTATTT_nat_bw1  2570765 139655 (5.4%)   no  450
S89_ACATTT_nat_bw1  1338242 39425 (2.9%)    no  378
S90_GGTGTT_nat_bw1  2634816 161102 (6.1%)   no  450
S91_AATGTT_nat_bw1  1121329 53323 (4.8%)    no  393
S92_GCGGTT_nat_bw1  2574415 176444 (6.9%)   no  472

Overall, only a small fraction (3~4%) of the reads were assigned to the transcript GTF, and I think these alignment fractions are too low. I wonder if you, as an expert in analysis of miRNA sequencing, can provide your comments on what would be a potential problem causing such low alignment fractions.

Thank you in advance,

Hoon

[Eduardo Andres Leon]

Dear Hoon, I see that you are already an expert on miARma ;). I hope you find it useful and easy-to-use.

Regarding your question, you only have to think if your results make sense. Although you have change the gtf file that doesn't mean that you have to find anything, what I mean is that in some cases the wet part remove bigger fragments. For example in miRNAs, the protocol needs fragmentation and sonication to remove big fragments. Thus for lnc-RNAs is better to use a RNASeq protocol rather to a miRNASeq protocol.

As a test, you can change the strand (now you have strand=no, try strand=yes), to make sure it is not a quantification problem Another way to check it (visually), is to use the IGV software using a bam file from Bowtie1_results folder and your gtf file. In such a way you can see the number of reads in each small mouse transcript

Those are my suggestions

Regards

[Eduardo Andres Leon]

Dear Hoon should I close the issue ?

Regards

[Hoon Kim]

Sorry, I should have done it. Please, close it. Thank you,

HK