My reliability tool (MRT)

Cryo-EM methods can in many cases render a density map that seems reasonable, but is it true?. Can users trust the outcome of a given processing workflow?. In recent years validation and methods to asses the quality and reliability of the maps have received much attention (Rosenthal et al, 2015). “My reliability tool” aims to provide several map reliability methods (two in the current version) that may help users to assess the quality of their reconstruction without any software installation requirement. One of the methods provided by SWT is the one referred to as “Phantom in the noise” (Heymann, 2014) that compares the effect of signal and noise images in reconstructing a map (Figure 1) The second method provided by SWT assesses the angular assignment reliability between a set of particle projections and the map calculated from them (Vargas et al, 2016), rendering the corresponding cluster tendency chart (Figure 2)

Link: http://scipion.cnb.csic.es/m/mypval

Tutorial

Embedded help guiding our users through the items to click (steps) to get reliability charts and data about their reconstruction providing already data and finalized results (3 different datasets).

http://scipion.cnb.csic.es/m/p_content?p=betagal

Workflow

This tool needs a set (around 4000) of aligned particles, and the volume reconstructed from those particles. None of the methods need high resolution data, so 64 pixels side images should be enough. First steps are the import steps for the volume and the particles. Once finished you can run the reliability protocols straight away.

VolumeAligned particles

SWT_MRT_W

Output of alignment reliability

A chart with information about the P value (comparison between the particle and noise alignment precision) for each particle. A value higher that 1 means that the particle aligns with more precision than noise.

overfittingchartbetagal

Example of an acceptable result

Output of BSOFT/Xmipp – validate overfitting

A chart with two series related to the resolution of the reconstructed maps; one from aligned particles and another from aligned noise. Comparing them will help you to see if your aligned particles may reconstruct a valid map or not.

Example of a probably non-acceptable result

Example of a probably non-acceptable result