3 and 4 was performed at 25°C on a 600 MHz Bruker Avance III spectrometer with a z-gradient triple resonance room temperature probe.

One-dimensional cross-sections through the 13 2D projections of a 4D APSY-HNCOCA experiment recorded as described in the text and in the caption to Fig.  3 .

The cross-sections were taken at ω ₄ ( ¹ H) = 8.957 ppm, which is the chemical shift of the peak identified in Fig.  3 with an arrow .

With the selected short measuring time, the intensity of the weakest NMR signals in the projections is comparable to the intensity of the thermal noise (Figs.  3 , 4 ).

In the experiment of Figs.  3 and 4 , the simultaneous analysis of the complete set of 13 2D projections thus enabled reliable recognition of the correlation peaks on the background of the thermal noise..

For example, in the experiment of Figs.  3 and 4 , 13 projections were sufficient to yield a good separation of correlation signals and artifacts, so that GAPRO could generate a complete peak list with 110 peaks.

Figure  9 shows the variation of the peak list resulting from the 4D APSY-HNCOCA experiment of Figs.  3 and 4 when different values of S_min are used.

For example, the values r_min  ≥ 5.0 Hz and Δ ν_min  ≥ 2.4 Hz yielded a complete peak list for the experiment of Figs.  3 and 4 (Fig.  10 ), which was recorded with a digital resolution of 4.7 Hz in the direct dimension, and digital resolutions between 5.6 and 15 Hz in the indirect dimensions, depending on the projection angles used for the individual 2D recordings. .

Application of the newly introduced GAPRO secondary peak filter had no impact on the final result obtained with S_min values in the range 5–8, which yielded a complete, artifact-free peak list from the 4D APSY-HNCOCA experiment of Figs.  3 and 4 (Fig.  9 ).

A1) were used to obtain backbone resonance assignments for TM1290, which is the same protein as was studied in the experiment of Figs.  3 and 4 .