FR and RF with abyss-pe

[GM]

Hi,

I'm trying to assemble a stranded PE Illumina RNA-seq using abyss-pe, and I was trying to understand which reads would be called /1 and /2. The protocol used was dUTP so I should be getting reads in the RF orientation (1st reads=reverse, 2nd reads=forward orientation). So, should the first read be labelled /1 and second read be labelled /2?

I did a small expt by taking only one set of PE reads. For one abyss-pe run, I labeled the first read /1 and the second read /2, while the for second run, I labeled the reads vice-versa. However, in both cases, abyss-pe says:

Mateless          0

Unaligned    838861  4.58%

Singleton   1459513  7.96%

FR         12094756  66%

RF              944  0.00515%

FF             1485  0.0081%

Different   3938969  21.5%

Total      18334528

FR Stats mean: 224.8 median: 193 sd: 113.6 n: 12093372 min: ../55 max: 1083

   ▄▇███▇▇▆▅▄▄▃▃▂▂▂▁▁▁______

Mate orientation FR: 12094756 (100%) RF: 944 (0.0078%)

The library XX.hist is oriented forward-reverse (FR).

In both cases, the orientation of the PE library is shown as FR. This is a bit confusing, since the # of contigs produced are very similar in both cases. Does the /1 and /2 labeling not matter for abyss-pe? Or am I doing something wrong?

Thanks

 

[Tony Raymond]

Hi,

I'm happy to hear you are trying out the stranded functionality of ABySS. Just to confirm, you are using abyss-1.5.1, and the 'ss=1' parameter on the abyss-pe command line, correct? In this case, ABySS assumes that the reads are oriented as \1 is reversed, and \2 being correctly oriented WRT the transcript (as in your case). So you shouldn't need to reformat your reads.

The statistics you list are fragment orientation statistics and should be read as:

FR : inward pointing - PET libraries

RF : outward pointing - MPET libraries

FF : same sense - ??? libraries

These have nothing to do with the numbers of reads that are correctly aligned.

Cheers,

Tony

--
You received this message because you are subscribed to the Google Groups "ABySS" group.
To unsubscribe from this group and stop receiving emails from it, send an email to abyss-users...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[GM]

Hi Tony,

I'm using 1.3.6 with the option SS=--SS as specified on the documentation page here: https://github.com/bcgsc/abyss#abyss. I should switch to 1.5.1.

If the fragment orientations do not indicate numbers of correctly aligned reads, is there a way to obtain that number via any of the intermediate files or the final output?

Thanks

[Tony Raymond]

Hi,

The 'SS=--SS' or 'ss=1' parameters were only implemented in 1.5.0 and later, so you do need to upgrade.

There are some numbers printed about # reads that map sub-optimally (could map better to the opposite strand), and # of reads that don't map but could if they were reverse complemented. To get these, you'll need to specify the abyss-pe parameter 'v=-v', and then you'll get lines like these after abyss-map runs:

Mapped 6131607 of 7626834 reads (80.4%)

Mapped 5970393 of 7626834 reads uniquely (78.3%)

Mapped 296160 (3.88%) reads to the opposite strand of the optimal mapping.

Made 41428 (0.543%) unmapped suboptimal decisions.

I could probably word this log output better, but at least you get the numbers you're looking for.

Cheers,

Tony