1. NMRbox Data-as-a-Service Overview
data archival and retrieval
software integration
data interchange
Synergy between BMRB & CONNJUR. BMBR handles data archival & retrieval (among other things). CONNJUR’s goal is software integration. They have in common the task of data management & interchange.
Projects
Analysis-as-a-service

2. Objectives 2 1 3
1. CONNJUR: capture metadata to save the state of NMR study.
2. CONNJUR as a deposition engine to BMRB.
3. M2M communication services between NMRbox and BMRB.
The four aims of TRD2 & how they (a) are related and (b) unify the missions of CONNJUR & BMRB.

3. Approach: CONNJUR Workflow Builder Spectrum Translator
Graphical software integration platform for spectral reconstruction
Spectrum Translator
Command-line tool for translating time and frequency domain data.
Integral component of Workflow Builder.
Sparky “R” Extension
Annotation for reproducibility
NMR-STAR Parser
Translation tool
CONNJUR Database
MySQL database managing datasets used by Workflow Builder

4. Approach: BMRB Application Program Interface (API)
Allows for software access to the BMRB database, both for data retrieval and deposition
Data Format Translators
CONNJUR, NMR-STAR, XML, JSON, NEX
Data Analysis & Visualization
DEVise visualization tool, Libraries in
R language, Validation tools
Deposition Engine
CONNJUR integration, automatic gathering
and deposition of data and important meta-data, including workflow specs

5. Workflow Builder

6. Time-domain and other files
Approach: NMRbox M2M data exchange
API
Query
response
BMRB
servers
Auto-query
generator
NMRbox
user
CONNJUR
database
CONNJUR
data harvester
Time-domain and other files
Spectral
processing
Peak lists
Auto assignments
Restraints
Structure
models
NMR
spectrometer
NMRPipe
Sparky
ABACUS
TALOS+
CNS

7. Time-domain and other files
Content Harvesting for Deposition
BMRB
Deposition
constructor
API
NMRbox
user
wwPDB
CONNJUR
data harvester
DRCC
Time-domain and other files
Spectral
processing
Peak lists
Auto assignments
Restraints
Structure
models
NMR
spectrometer
NMRPipe
Sparky
ABACUS
TALOS+
CNS
CONNJUR workflow manager

8. NMRbox/CONNJUR Deposition Service
Dynamics
Chemistry
Interactions
NMR-STAR
Raw data
Spectral data
Derived data
Data annotation
CONNJUR
Structure
&
related data
Metabolomics
results

9. NMR & supplemental data
Approach: NMRbox Data Mining – BMRB Archive Content
Metadata
chemical structure, natural source, sample, experimental detail
Imported data
coordinates, restraints,
phi-psi angles
Validation results
LACS, AVS, PANAV, SPARTA+, CING, MolProbity
Biological
NMR & supplemental data
Derived data
back calculated chemical shifts,
BLAST alignments
Data interpretation
citations
External data links
PDB, UniProt, KEGG, PubChem

10. Approach: NMRbox BMRB Data Mining
Exploring the BMRB archive for new knowledge
Expose the BMRB relational database and additional value added data for query and analysis from within the NMRbox platform
Develop information search and analysis tools that encompass the breadth of the BMRB archive
Brief general examples
Prediction and analysis of intrinsically disordered protein conformational space from NMR spectral parameters and derived data
Search for links between NMR parameters, low population biopolymer conformers, and biopolymer interactions with other biopolymers and ligands
Extract RNA chemical shifts and statistics for improving automated chemical shift assignment methods and structure analysis
Integration of molecular dynamics simulations with NMR experimental results to understand biopolymer conformational sampling

11. Data mining and visualization on BMRB – R libraries
CA-CB
Chemical shift
Distibution in
BMRB per
residue

12. Data mining and visualization on BMRB – R libraries
Comparing
HSQC spectra
for homologous
entries

13. Data mining and visualization on BMRB – DEVise
Comparing
HSQC spectra
for homologous
entries

14. Impacts (CONNJUR)
1- Additional metadata is critical to foster reproducibility.
It serves dual purpose of allowing us to populate new instances of NMRbox.
2- Eases the burden on the NMR community for submitting data to the BMRB. As CONNJUR is capable of tracking larger amounts of intricate data than the spectroscopist is likely to be willing to provide – the BMRB depositions will be fuller.

15. Impacts (BMRB)
1 - BMRB content relevant to the NMRbox users, and possibly unknown to them, will be exposed and presented without the need for user knowledge of the BMRB archive architecture or content or user training.
2 – New possibly unexpected correlations between NMRbox user data and the full BMRB archive (experimental, derived and/or predicted, validation, and other kinds of data) will be advanced.
3 – Workflow and preservation meta-data archived for reproducibility.

16. Thank you!
Any questions?

17. Data mining and visualization on BMRB – R libraries
TOCSY
EXAMPLE

18. Personnel UConn Health Wisconsin Admin Infra Train Dissem CS DBPs TRD1
Hoch
Maciejewski
Schuyler
Gryk
Ulrich
Eghbalnia
Gilman
Gorbatyuk
Moraru
Livny
Maziuk
TBN
TBN1
TBN2
TBN3
TBN4
TBN5
UConn Health
Wisconsin

19. Metadata Examples for M2M and Data Mining
Applications
Biopolymer sequence, natural source including location
Mining
Intermediate data (restraints, chemical shifts, peak lists)
Value added data (secondary structure elements, physical properties, etc.)
Sample conditions (pH, temperature, pressure, ionic strength)
Selection
Validation report content
User process annotations
Best practices
Software application parameter files
Pulse programs
Spectrometer field strength
Sample contents (buffers, salts, stabilizing agents, others)
Author names
Keywords
Descriptive
User text annotations

20. Personnel Personnel Effort Role Gryk 2.4 Co-leader of TRD2
Extend CONNJUR data model
Ulrich
0.84
Livny
0.24
Collaborator – systems design
TBN1
9.6
Application architect
CONNJUR software components
Query Engine design
Maziuk
1.2
Systems administration
TBN3
8.4
Researcher/programmer
BMRB software components
TBN5 6
Programmer

21. CONNJUR Schema Expansion (Aim 2.1)
Current CONNJUR
strengths
Spectrometers
Pulse programs
Parameters
Output data
Processing software
Fully extended CONNJUR
schema
Current NMR-STAR
strengths
Citation
Molecular system
Sample
Conditions
Spectral data
Derived data
Current NEF
strengths
Structure software
Input restraints data
parameters

22. NMR Computational Pipeline1 2 3 4 +
L10 A5
< 5Ǻ
Four broad phases of computation. 1st is on spectrometer – we don’t touch that. 2nd is handled by CWB. 3rd & 4th is the realm of peak lists, resonance, spin systems – semi-automated peak pickers, assignment, NOE assignment & structure determination.
Spectrometer
Acquisition
Spectral
Reconstruction
Spectral
Analysis
Biophysical
Characterization