Two new papers published today in the Matters Arising section of Nature Methods discuss the power and limitations of DENSS. Drs. Petr Konarev and Dmitri Svergun, pioneers in the field of SAXS data analysis and modeling and authors of the highly popular ATSAS package, published a Matters Arising article discussing their views on the accuracy and limitations of DENSS and its ability to reconstruct 3D density ab initio from 1D scattering profiles. As for all Matters Arising articles, I was given the opportunity to respond to their critiques and published my response alongside their article. The Konarev and Svergun article can be found here or here and my response can be found here or here.
As a brief summary, Konarev and Svergun explain that their investigations showed that DENSS can indeed reconstruct accurate ab initio shape reconstructions from SAXS data and is thus an important contribution to the field. However, they hold the view that DENSS cannot reconstruct actual density accurately as is claimed in my original article (just envelopes). They also claim that the averaging procedure used by DENSS results in a low-pass filter effect that results in artifacts, namely that density in the interior of the particle is not real.
In my response, I present new reconstructions (Figure 1) proving that DENSS can indeed reconstruct actual density, not just envelopes, provided sufficient contrast difference exists between components of a multi-component particle. In particular, I show that the Membrane mode of DENSS can accurately recapitulate particles with negative contrast (using experimental SAXS data from DDM micelles*), which is not possible with conventional bead modeling algorithms, and that the reconstruction from an artificially simulated DNA-protein complex (where the DNA contribution was doubled) showed significantly greater density for the DNA molecule as expected. I also demonstrate that the averaging procedure in DENSS is an appropriate way to improve the resolution of individual reconstructions through analysis of the Fourier Shell Correlation with a known protein structure.
Academic discourse such as in the Matters Arising articles published today is an important part of the scientific process and ensures that thorough validation of new methods and findings is presented to the community. I hope that these articles showcase the power of DENSS as well as the dangers of over-interpreting scientific results.
*SAXS data of DDM micelles kindly provided by Frank Gabel. Details in my response article.