RNA editing with GNUMAP

[W. Evan Johnson]

Brent,

Are you sure it is not working? Are you accounting for the fact that GNUMAP reports the positive strand bases, and therefore you are looking for T->C changes when you are looking for negative strand editing? I guess this depends on whether you are using .sgrex or .sam for your analysis.

If this is not the problem, then Nathan can you address this issue ASAP? I think this is the same problem we are having with our bisulfite mapping.

Thanks!

Evan

On Mar 3, 2011, at 2:44 PM, Brent Shepherd wrote:

> Hi Bryan,
>
> so it appears that the GNUMAP issue is still an issue (the minus strand part, anyway). I ran my code using the updated version, and I didn't get anything significant. So, I've reverted back to the older method and am in the process of that. That should hopefully be done by Monday. I'll keep you posted--I just wanted to give you the update.
>
> Brent

[Brent Shepherd]

Evan and Nathan,

I've accounted for T-C changes and have made the necessary adjustments. Upon looking into it further, it seems that everything is mapping fairly well (as in how it should be), except the minus strand isn't catching the editing sites. For example, in mir-800, I was mapping 28 reads in our "tweaked" method, and I'm only mapping 24 with the minus strand flag. In other words, the four that were mapping to the other strand with editing aren't being included in this mapping.

After thinking about it, could this possibly be because GNUMAP is looking for A-G editing still, when it should technically be looking for T-C mapping on the minus strand? This is my best guess. If this were the case, then the PWM would have to be ... transposed? ... in order to properly account for strand mapping.

Is this a possibility? The read counts matched are quite similar, except for the edited ones...

Brent

[Brent Shepherd]

Let me rephrase a few of those sentences, as I'm not sure it was clear...

when I manually inverted the genome and mapped it backwards using --plus_strand mapping, I'm getting nearly the same results as the new --minus_strand mapping included in the latest version. The only difference appears to be that the edited strands aren't mapping (hence the 24 vs. 28 mapped to mir-800) with the --minus_strand flag. I think this is because the PWM needs to be transposed. The -d option that allows for A to G editing (and this would be the same issue for the bisulfite sequencing) would need to effectively account for T to C editing on the minus strand. I hope this adds a little clarity to my previous e-mail...

Brent

[Nathan Clement]

I think you are correct.  This is a problem with bisulfite mapping.  When we try and map the negative strand, instead of allowing a T to transition to a C, it's allowing A's to transition to G's (because we simply reverse-compliment the reads).  I'll need to make a few more changes to this to bring it up to speed.

Thanks,

Nathan

[W. Evan Johnson]

Nathan, thanks for doing this promptly. We need results from both editing and bisulfite ASAP!