Transcript Data Support for Life Sciences

Data Management Support
for Life Sciences
or
What can we do
for the Life Sciences?
Mourad Ouzzani
[email protected]
The Big Picture
Proteomics, Metabolomics, & Cytomics
Sample
Data
Purdue BioScience Pipeline
Experiments
Storage
System
Data
Streaming
System
Samples
Processed Data
Raw MS
file
Peak Deconvolution
Noise
filtering
Online
Analysis
Raw
peak
group
Peak group
selection
Single
scan
clusters
Doublet
detection
Charge
fitting
Chromatographic
refinement
Mixed
doublet
rescue
Multiscan
clusters
Isotope
fitting
Data
Mining
Priority
list
Protein
identification
Calculation of
Differentials
Web
Service
Internet
Biological Databases
Databases
Integration
System
Raw Data
Intelligent Instrument Control
Protein Identification
Mass
Spectrometer
Diseased
Sample
Samples Preparation
& Transformation
(LC, isotoping, etc.…)
Non Diseased
Sample
Mass
Spectra
Data
Analysis
Step 1
Step 2
Step 3
Data
Mining
Massive Data Storage
Issues




Time sensitive data
Limited sample quantities
Experiments repetition
Massive data
Intelligent Instrument Control
Mass
Spectrometer (1)
Instrument
Vendor PC (2)
Instrument
Intelligence (3)
Samples
Network
Databases (4)
Raw Data Archives (5)
Benefits



The outcome of IIC will be biological
knowledge instead of raw mass spectra.
The biological knowledge is backed up
by data acquired by IIC.
Scientists do not need to review the raw
mass spectra.
Data Flow in IIC
Nile Support and others
IIC Issues




IIC system development
Non-proprietary API for both data
collection and control of the instrument
Optimized storage for Massive data
(Instrument Output and Sequences)
etc.
Data Stream Issues






Data filters that identify interesting data and
reduce chemical noise
Algorithms for rapid identification of the base
peaks and the number of peaks in the
spectrum
Algorithms for prediction of upcoming peaks
Online statistical analysis over the streams
Data summaries on different granularities
etc.
Data Integration
Non-glycosylated peptide identification
MS/MS Spectra
preprocessing
de novo sequencing
APLIXYX
CLIKWDYR
database search
stats auto-validation
protein validation
Protein List
Data Integration and Informatics
Web Browser
Web Service
Consumer
Web Service
Invocation (SOAP)
Queries
Informatics
Toolbox
NON-GLYCOSYLATED
PEPTIDE
IDENTIFICATION
Web
Service
Access
GLYCOSYLATED
PEPTIDE
IDENTIFICATION
Web Service
Description (WSDL)
Metadata
Repository
Database
Discovery
Database
Locator
Request Handler
Query
Optimizer
Execution
Engine
WebGlyco
Manager
Mapping
Agent
Wrappers
Biological
Databases
Glycoprotein Databases
Other Protein Databases
Data Integration Issues







Databases description and organization
Schemas mediation
Annotation and Provenance
Use of model management techniques
Query processing and optimization
Web-service access
Implementation and deployment
Requirements








Data types diversity: sequences, graphs, 3D
structures, etc.
Unconventional queries: similarity, pattern
matching, etc.
Uncertainty (probability)
Data curation: cleaning and annotation
Data provenance (pedigree)
Large scale: 100s of DBs
Terminology management (semantics)
etc.
Data Correlation



Non-overlapping Schemas (different
instruments or scales of resolution)
Contradictory information (experiments
with different assumptions)
Comparing data only after matching
their context (constraints)
Other Issues

?
IIC Information Flow
sample
Step 1
Step 2
N
Interesting ions?
Y
Priority list of interesting ions
Empty priority list?
N
Step 3
N
N
QA/QC?
Peptide identification
Y
Protein identification
External Databases query
Y
Intelligent Instrument Control

Algorithms design






Spectra Deconvolution
Online analysis (protein/peptide identification)
Online peaks Identification for feedback
Data filters and noise removal
Prediction of upcoming peaks
Experimental Simulation


In silico generation of spectrum
Algorithms simulation
Intelligent Instrument Control

Experimental settings


Selection of a biology system, e.g., yeast
Two types of experiments



Target analysis
Global analysis
Integration with the instrument




Data collection
Control of the instrument
API
Actual implementation (algorithms)
Intelligent Instrument Control


Online data mining
Other Issues:


Optimized storage of massive data
Data representation (streams, database)
Integrated Access to Glycoprotein Databases

Informatics tools





Glycosylated peptide identification
Non-glycosylated peptide identification
Enabling uniform access to different
glycoprotein databases
Databases description and organization
Schema mediation
Integrated Access to Glycoprotein Databases


Query Processing
Data correlation





Non-overlapping schemas
Contradictory information
Sequence alignment
Web service enabled access
Target databases selection (focus)