Hello,
Tim will have to look into this. Another alternative way is to do the
sequential linking before you hit the "Add HA/CO/CA/CB" button (that's the
way Tim designed the system). Tim can see that doing it the other way
around might cause problems, including duplicate resonances, except that
this is something you might want to do. Note that the chemical shift
information, although not quite as good, is still used irrespective of
whether you have added the atom types.
Wayne
On Tue, 18 Jan 2011, Aswani Kumar wrote:
> Hi
>
> After the recent update of the Analysis, I have included the experiments Hc_ccoNH, 1H,15N-HSQC-TOCSY, HNHA and HNHB in the Sequence Assignment System besides the usual carbon correlation spectra.
>
> I have the following set up for Sequence Specific Assignments:
>
> Query Spectra: HNCO, CBCACONH (in window:2), Hc_ccoNH (window:3)
>
> Target Spectra: HNCACO, HNCACB (window:4), 1H,15N HSQC-TOCSY, HNHA, HNHB( window:5)
>
> Addition of proton correlation spectra in the sequence specific assignment system has improved the confidence of assignments. However I have the following problem.
>
> After setting the sequence link in the pop-up " M::Assignment::Protein Sequence Assignment ", when I click on "Add HA/CO/CA/CB", Analysis correctly identifies the CA, CB and CO resonances of the residue i. It also correctly assigns the HA in HNHA spectrum. Moreover, it also links the side chain protons of the residue "i" correlated to the Amide Proton of residue (i+1). However it assumes that all those proton resonances are new resonances (for instance the HA of residue "i" is again initialized as a new resonance in Hc_ccoNH spectrum). As a result there are more proton resonances linked to the residue type (for eg., two proton resonances with alpha proton chemical shift and so on). Hence the residue type prediction is going haywire completely. For example, an aspartate is predicted as a Lysine or Arginine (i.e prediction is biased towards residues with larger number of side chain protons). It is also reflected in the table on the right bottom of the pop-up " M::Assignment::Protein Sequence Assignment " as well as when we use the "Predict Type" feature. As a result the the suggested sequence locations (bottom left) and residue types (bottom right) are completely wrong.
>
>
> The way I try to get the Amino acid type predicted correctly by Analysis is to merge the resonances or remove them from the spin system. Merging is possible only when I am sure of the assignment; and removing them from the spin system is repetitive work as they actually belong to the amino acid system and need to be reassigned later.
>
> How can I get over this issue? Basically the proton resonances for the side chain protons are initialized more than once. First time on the HNHA spectrum and again on the the Hc_ccoNH spectrum. This is resulting in bad prediction of the residue type. As an immediate rescue measure, is it possible to instruct Analysis not to use the side chain proton chemical shifts for predicting the residue type?
>
> Regards
> Aswani Kumar
>
|