I have come up with couple of changes that hopefully sorts the issue.
Firstly, I have added an extra check to ensure that assignment propagation
never unaliases a peak to outside the spectrum limits; observing the sweep
width or the min/max aliased frequencies if set. Thus, setting narrower
spectrum limits can be used to stop unwanted aliasing attempts.
Secondly, I have added an internal toggle so that the user will be
prompted before propagation causes an aliasing change. This warning is
only made for the manually invoked propagation, so doesn't affect more
automated routines. Naturally, if lots of unaliasing is really required it
is now best to sort that first before attempting assignment, otherwise
there will be several confirmations required.
We can make the warning subject to a profile setting if the new situation
becomes too cumbersome.
On Tue, 16 Mar 2010, Andrew Fowler wrote:
> I'm not sure why I haven't reported this before, but Mattias reporting an
> unrelated aliasing bug jogged my memory.
> When I run sidechain proton experiments (i.e. HBHAcoNH, HccoNH-TOCSY) I
> usually shift my indirect proton carrier to 3ppm to maximize digital
> resolution in that dimension. This is I believe the key to the issue, so I'm
> not sure how many other people get hit by it - Varian BioPack for sure and I
> think Bruker stock experiments leave the carrier on water.
> In any case, I typically already have the backbone amides assigned before I
> start looking at these experiments, so I'll pick the indirect proton peaks,
> select them along with the corresponding peak in the 15N-HSQC, then
> M->Assign->Propagate Assignments to assign the 15N and HN (direct) shifts.
> In the case where one of the indirect proton shifts happens to be within
> tolerance of where the amide shift would be if it were aliased (which seems
> to happen several times per experiment), Analysis automatically but
> incorrectly assumes the peak is aliased, sets it that way, and assigns the
> sidechain proton to be HN. I then need to remove the assignment for that
> dimension, manually unalias the peak, and reassign it to the correct proton.
> Is it possible to either turn off this "feature", set some sort of "never
> alias" flag for a given spectrum, or have a prompt to confirm before
> automatically aliasing a shift?