# Extending MolProbity Diagnosis & Healing Methods to Empower Better CryoEM & Xray Models at 2.5-4A Resolution, plus Versioned, Redeposited "GEMS" for Important Individual Structures

> **NIH NIH R35** · DUKE UNIVERSITY · 2021 · $389,399

## Abstract

Summary/Abstract
 My lab's MolProbity web service for diagnosing and guiding correction of local modeling errors in
macromolecular crystal structures has proven highly effective and is used by most of that research
community worldwide. However, as urgent need has recently arisen for new methods that can extend
MolProbity's reach to poorer resolutions in the 2.5-4Å range. That applies to the exciting, big "molecular
machines" solved by crystallography, and especially to the surge of "high resolution" cryoEM structures
now accessible since the revolution in detector and image collection technology. For several reasons, our
existing diagnostic criteria (and everyone else's) break down at resolutions where peptide orientation is no
longer visible in the map density. We have developed a new tool called CaBLAM at a multi-residue length
scale, which proved extremely effective in our assessments for the CryoEM Model Challenge. That
experience also gave us ideas for several other criteria that could diagnose errors common at 2.5-4Å,
such as flipped-over peptides and local sequence misalignments. This proposal aims to create a unified
toolkit for these resolutions in MolProbity, an urgently needed functionality which does not currently exist
anywhere. The more biomedically important these new structures are, the more crucial it is not to model
the wrong amino acid at the critical site for catalysis, dynamics, drug design, or other specific binding.
 When developing the original MolProbity site we found that such new tools never act quite as one
expects, and so extensive testing in real, varied production use is necessary. For this proposal, we will
make those tests immediately valuable to the end-users of structures by a workflow called "GEMS". We
will identify individual x-ray or cryoEM structures that are a most-used archetype for their molecule but
which contain local misfittings that affect their functional interpretation. If the depositor-of-record is willing,
we will collaborate on healing those problems using the new tools. An important change this year to
facilitate this goal is that the wwPDB is implementing an archival versioning system to enable coordinate
updates without changing the PDB code, which makes structural biologists much more willing to deposit a
corrected GEM structure. The lessons from creating these GEMS will be a major guide in tuning the new
toolkit for maximum effectiveness, ease of use, and a minimum of false changes at the lower resolutions.
 There is now a new threat to MolProbity's ability to provide complex, dynamic services open to the
worldwide scientific community in an increasingly hostile internet environment. MolProbity has recently
been the target of extremely sophisticated attacks, which have required very significant personnel effort
including near-constant monitoring, reinstallations, and close collaboration with medical center security
experts. We have plugged some vulnerabilities but more work is needed, such a...

## Key facts

- **NIH application ID:** 10170382
- **Project number:** 5R35GM131883-03
- **Recipient organization:** DUKE UNIVERSITY
- **Principal Investigator:** DAVID Claude RICHARDSON
- **Activity code:** R35 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2021
- **Award amount:** $389,399
- **Award type:** 5
- **Project period:** 2019-06-01 → 2024-05-31

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/10170382

## Citation

> US National Institutes of Health, RePORTER application 10170382, Extending MolProbity Diagnosis & Healing Methods to Empower Better CryoEM & Xray Models at 2.5-4A Resolution, plus Versioned, Redeposited "GEMS" for Important Individual Structures (5R35GM131883-03). Retrieved via AI Analytics 2026-05-23 from https://api.ai-analytics.org/grant/nih/10170382. Licensed CC0.

---

*[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*