Chazin Wisdom: Automated tools for NMR

I. Peak picking:

AUTPOSY

Cost: Free

OS: SGI

Ease of use: Moderate

This is a fairly straightforward package that can automatically pick peaks in multidimensional spectra. I haven’t played around with it that much; the procedure is basically to load the spectrum and execute one of the provided scripts for doing the peak picking. Can take a significant amount of time for 3D spectra. Peak lists can be exported to XEASY format, although there is an extra spacebar character at the end of the first line. (This character must be removed in order to import the peak list into NMRView.) This is the easiest and probably least effective use of the software, but I found the peak picking to be very good.

Sparky

Cost: Free

OS: Windows, Linux, Mac

Ease of use: Easy to Moderate

Though this software does a lot more than peak picking, I wanted to mention that in this capacity (automated peak picking), Sparky does not do a very good job. There is no easy way to auto pick the entire spectrum (that I know of yet), and keeping track of which peaks have been picked in 3D spectra can be difficult.

II. Backbone assignment:

PACES

Cost: Free

OS: Windows (requires Excel)

Ease of use: Easy

This package has some significant advantages and disadvantages. The major criticism is that you have to assemble the spin systems in advance, and the facility for importing spin systems from XEASY peak lists was not very effective for in my test run.

Each spin system is a single peak on the 15N-1H HSQC spectrum. Peak pick this 2D spectrum and write an XEASY list. Import it into PACES to generate all your spin systems. Now you can go through your HNCA / HN(CO)CA / HNCACB / HN(CO)CACB spectra to enter the CA, CA(i-1), CB, CB (i-1) resonances associated with each HN,N pair. You can also enter your HNCO, HA, and HB data here. Once this data is entered, however (I went back and forth between Sparky and Excel to assemble the spin systems), the software appears to be fairly effective at correctly identifying high probability stretches of assignments. With an incomplete data set (many peaks missing from the HNCACB) the program correctly assigned about a third of CRCC. I will test the program further with better data.