The MoBIoS Project Molecular Biological Information System

1
The MoBIoS ProjectMolecular Biological
Information System

• Daniel P. Miranker
• Dept. of Computer Sciences
• Center for Computational Biology and
  Bioinformatics
• University of Texas

Weijia Xu, Rui Mao, Will Briggs, Smriti
Ramakrishnan, Shu Wang, Lulu Zhang
2

• ProblemIn Life Sciencses, database management
  systems (DBMS) serve as glorified file managers.
• Little use of sophisticated data and
  pattern-based retrieval
• Real scientific and technological problems

3
When biological data is put in to an RDBMS

• Primary data is stored in text or blob fields
• Annotations may be relational
• Data retrieval
• Filter DB, sequential dump, O(n), to utilities
• E.g. BLAST,

Organism Function Sequence
Yeast membrane AACCGGTTT
Yeast mitosis TATCGAAA
E. Coli membrane AGGCCTA
4
Linear Data Scans, O(n), Endemic in Life Sciences

• Sequences
• DNA, RNA, Protein databases
• Mass Spectra
• proteomics
• Small Molecules Protein Structure
• Protein interaction
• Rational drug design
• Pathways (graphs)
• Phylogenies (graphs, trees in particular)

5
Scope To Find Common Ground Both Biology and
DBMS Have to Move
DBMS
Biological Information System
Metric-Space Database as the Common Ground
6
Metric Space is

• a pair, M(D,d),
• where
• D is a set of points
• d is metric distance function with the
  following properties
• d(x,y) d (y,x)
  (symmetry)
• d(x, y) gt 0, d(x,x) 0
  (non negativity)
• d(x,z) lt d(x,y) d(y,z)
  (triangle inequality)

x y z
7
Definition - By Analogy

• A Spatial Database Management System
• Extend relational DBMS
• Special indexes for 2D and 3D data k-d and
  R-trees
• New data types
• Geographic information systems
• Topographic maps
• Buildings and the like

• A Metric-Space Database Management System
• Extend Relational DBMS
• Special indexes for metric-spaces
• New data types
• Biological information system
• Life science data types

8
Develop index structures to support distance
nearest-neighbor queries

• Well studied in main-memory
• But by no means a closed problem
• In databases (external/disk based methods)
• Embryonic
• Many myths
• Often assumed to be the basis of multimedia
  database systems

9
How to build a metric-space index

• Three algorithmic classes Tasan, Ozsoyoglu 04
• Vantage points
• Hyperplanes
• Bounding spheres

10
Vantage Point Method BurkhardKeller73
11
Vantage Point Method
Choose a point,VP
And a radius, R
12
Vantage Point Method

• Given VP, R
• The predicates
• d(VP,x) lt R
• d(VP,x) ? R
• Divide the set into two equal halves
• apply recursively

Choose a point,VP
And a radius,R
13
Query, q, range r
r
q
14
Query, q, range r

• if
• d(q,VP) gt R r
• then
• all neighbors are outside the sphere

VP
R
r
q
15
Multi-vantage point method
16
Multi-vantage point method

• Consider d(VPi, x) a projection onto an axis
• Looks like a k-d tree
• Choose number k d

17
Myths

• Solved problem M-trees Ciaccia et.al. 96, 97
• I cant get them to work on anything but their
  original synthetic data generator
• Good choice for vantage points is to find
  cornersYianilos93 (farthest-first clustering)
• Might be true for euclidean spaces
• Early result, not true for our data
• High dimensional indexing always asymptotically
  reduces to linear scans.
• Formal result based on an assumption of uniform
  data distributions.

18
Comparison of Three Methods of Metric-Space
Indexing
19
Open problems

• Is there a general metric-space index structure
  that is generally good for most work loads.
• We are optimistic mvp trees further tuning
  will be a useful answer
• Hyperplane methods are fair game there is
  circumstantial evidence that that is key
  component in Googles search engine.
• No work addresses clustering data pages on disk.
• Metric-space join algorithms

20
Biological Models are Usually Based on Similarity

• Similarity
• Biologist like scoring functions that reward each
  similar feature with a positive number
• Intuitive
• Distance
• More Similar ? smaller numbers
• Identical ? 0

21
But Do Metric Models Capture Biology?

• Metrics are a subset of possible mathematical
  models

.
22
Sequence Problem 1

• Sequence similarity based on weighted edit
  distance
• Accepted weight matrices, PAM BLOSSUM, are not
  metric
• Log-odd matrices negative values
• Defy simple algebraic normalizationTaylorJones93,
  Linialetal97

23
Our First Result mPAM XuMiranker04

• Dayhoffetals PAM Derivation74
• Took a set of closely related protein sequences
• Developed a phylogenetic tree
• Counted substitutions to transform one sequence
  to another
• Tree determines a measure of time

24
PAM vs. mPAM t 1/f

• Using original substitution counts
• PAM frequency of substitution
• S(a,bt) log P(ba,t)/qb
• mPAM expected time between substitutions
• D(a,b) 1/log(1 ?(P(a,x)P(b,x))

x
25
Sequence Problem 2

• Sequences long units (identity for storage and
  retrieval)
• Genes
• Chromosomes
• Analysis comprises comparing small substrings

26
Soln Sequence View

• New view type
• Breaks sequences into q-grams

create SEQUENCEVIEW rice_sview as SELECT CREATE
FRAGMENTS (, 3, 1) FROM WHERE USING
HAMMING-DISTANCE
27
Materialize as an Index
D(AAA) 2
Rowd Offset Logical Fragment Logical Fragment Logical Fragment Logical Fragment Logical Fragment Logical Fragment
R1 1 A C A
R1 2 C A A
R1 3 A A C
R1 4 A C A

R2 1 A T C
R2 2 T C A
R2 3 C A A
R2 4 A A A

D(ACA) 1 D(CAA) 0 D(ATC) 1
Genomes Genomes
Rowid Seq
R1 CAACA
R2 ATCAAA
R3










28
Status

• Started with McKoi
• A Java open source object-relational DBMS
• (Think of Postgress written in Java)
• Added
• Biological data types
• Metric-space index
• Extending SQL engine (in progress)

29
Computed in MoBIoS

• Compare Arabidopsis Genome X Rice Genome
• Locate nucleotide patterns of form
• primer pair candidate
• Eliminate non-unique primer candidates
• Merge overlapping primer candidates
• Usual implementations O(n2), n 109

Rice Arab.
?18 Matching Nucleotides
?18 Matching Nucleotides
Rice Gap 400 3000 Long Arab. Gap 400 3000
Long
30
mSQL Query to locate candidate primer pairs

• SELECT merge(R1.fragment, A1.fragment)
• FROM
• G1_sview R1, G1_sview R2, G2_sview A1, G2_sview
  A2
• WHERE
• distance(HAMMINGDISTANCE', R1.fragment,
  A1.fragment) lt 1.0 AND distance(HAMMINGDISTANCE'
  , R2.fragment, A2.fragment) lt 1.0 AND
• (FRAGOFFSET(R2.fragment)-FRAGOFFSET(R1.fragment))
  gt 400 AND
• (FRAGOFFSET(R2.fragment)-FRAGOFFSET(R1.fragment))
  lt 3000 AND
• (FRAGOFFSET(A2.fragment)-FRAGOFFSET(A1.fragment))
  gt 400 AND
• (FRAGOFFSET(A2.fragment)-FRAGOFFSET(A1.fragment))
  lt 3000
• GROUP BY R1.fragment, A1.fragment

31
Query Plan

• Arab. Genome, O(n)
  Rice Genome, O(m)
• Offline Build Sequence
• View O(n log n)
• Compare O(mlogn)
• Indexed Nested Loop
• Eliminate Duplicates
• Eliminate Low Complexity
• Primers (LZ compression)
• Merge Overlapping Primers
• 10,000 conserved

32
Preliminary Results

• Found 13,418 possible primer pairs from MoBIoS
• 100 best candidates BLASTed for matches in
  GenBank
• 15 matched other plant genes and the primers
• At least 2 of 15 showed potential after PCR
  amplification against Helianthus and
  Phalaenopsis.

33
MoBIoS Architecture(Molecular Biological
Information System)
34
Analysing Mass-Spectra

• Spectrum Histogram of Mass/Charge Ratios of a
  collection peptides
• Similarity Shared peaks count Inner Product
• (0100101) (0111100) 2

35
Cosine Distance Approx. Inner Product

• Drs 1 xrxs/(xrxr)1/2(xsxs)1/2
• shown store and retrieve mass-spectra
• using cosine distance, and it scales

36
mSQL Query for Protein Identification by
Mass-Spec. Signature Database Look

• SELECT Prot.accesion_id, Prot.sequence
• FROM protein_sequences Prot, digested_sequences
  DS,
• mass_spectra MS
• WHERE
• MS.enzyme DS.enzyme E and
• Cosine_Distance(S, MS.spectrum, range1) and
• DS.accession_id MS.accession_id
  Prot.accesion_id and
• DS.ms_peak P and MPAM250(PS, DS.sequence,
  range2)

37
Matching Electrostatic Shape of Molecules
38
Still benefit from grid-services

• Intermittently, but regularly compile (recluster)
  the indices O(nlog n), n gt 106
• Rational drug design O(log n) finite element
  solutions to traverse search tree.
• Make a service call to the grid for these
  operations only
• Mirror data contents to minimize I/O
• Since need is intermittant, one grid serves many
  MoBIoS servers

recluster
G R I D




MoBIoS Server
New index
Shape match (FEM)
Distance(real)
High speed I/O
Mirror DB-Contents
39
Hyper-planes Ulhmann91

• If d(x,h1) lt d(x,h2) then x assigned to h1

h1
x
h2
40
Develop a Hierarchical Clustering
C
A
E
B
D
F

• Hierarchy of Bounding spheres, (center, radius),
• Bounding spheres may overlap
• Inspired by R-trees