Indeed, the code is using the “BMRB” “reference” data in data/ccp/nmr/NmrReference directory to determine the random coil values. I’ve asked Tim if he remembers where those come from.
Wayne
> On 8 Jan 2019, at 13:58, John Kirkpatrick <[log in to unmask]> wrote:
>
> Hi Niko,
>
> I was a bit curious about this, so I had a dig around. It seems that the random-coil shifts (and secondary-structure shifts) are coming from the DANGLE code, which is using random-coil shifts from J. Biomol. NMR 18, 43-48 (2000). These were measured in 8 M urea at pH 2, so I suspect the low pH accounts for the big discrepancies for Glu and Asp (I also see similar differences). For the proline Cb, it looks as if it is the random-coil shift for the cis geometry (34.8 ppm) that is being used rather than that for the trans (32.22 ppm); I guess this means there could be a bug somewhere, since both the cis and trans values are present in the reference table.
>
> Cheers
> John
>
>
> On 08/01/2019 05:20, Niko Loening wrote:
>> Can somebody explain to me what values are used as the baseline for
>> calculating the secondary chemical shifts for Cb's in the Secondary Structure
>> Chart? The references in the documentation only refer to Ca/Co/N/HN/Ha
>> chemical shifts. I've got a suspicion that the values that CCPN has, at least
>> for proline, are incorrect.
>> In my case, I've got a (mostly) disordered protein with proline Cb shifts
>> between 32.0 and 32.1 for the five prolines. These values are very close to
>> the random coil chemical shift values reported by Wishart (J. Bionmr, 5 (1995)
>> 67-81) and also very close to the BMRB average value (31.9 ppm). However, the
>> secondary chemical shift reported by CCPN analysis is a whopping -2.7 for
>> these residues. Other Cb shifts are not wildly off, so its not a referencing
>> problem on my part, and I'm very sure these prolines are in the trans
>> conformation based on their Cb and Cg shifts.
>> In fact, when I use CCPN to generate a set of predicted random coil chemical
>> shifts (Data Analysis -> Measurement Lists -> Make Random Coil Chemical Shift
>> List) I find that the resulting list seems to have some rather different
>> values for the following residues compared to the results published by Wishart
>> in 1995.
>> Asp (Co, Ca, and Cb) Glu (Cb and maybe Co) Gly (Co) Pro (Co and Cb)
>> There also might be deviations for Cys, Phe, Tyr, and Thr but my protein
>> doesn't have those amino acids so I couldn't check them easily. Of course,
>> things have progressed in the 20 years since Wishart published his results so
>> CCPN might be using newer data or a more sophisticated algorithm. But, based
>> on what I'm seeing, I still think there are some mistakes in CCPN's data.
>> So, what algorithm is CCPN using to predict random coil chemical shifts? Is
>> there a way to correct this?
>> Thanks,
>> Niko Loening
>> ########################################################################
>> To unsubscribe from the CCPNMR list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCPNMR&A=1
>
> ########################################################################
>
> To unsubscribe from the CCPNMR list, click the following link:
> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCPNMR&A=1
########################################################################
To unsubscribe from the CCPNMR list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCPNMR&A=1
|