Query sequence

1
Structure-Sequence alignment Structure is
better preserved than sequence
Query sequence MTYKLILNGKTKGETTTEAVD
AATAEKVFQYANDNGVDGEWTYTE
Me!
Me!
Me!
Me!
2
How can we match a sequence and a structure?
MTYKLILNGKTKGETTTEAVDAATAEKVFQYANDNGVDGEWTYTE
Sequence Similar Sequences take this structure
(but remember sequence is less preserved than
structure)
Pair-InteractionHow well do AAs get along
(Positive hate positive? Maybe not?)

• more
• 2nd structures prediction.
• 2nd structures constraints (ß-strands forming ß
  -sheets)
• etc.

Solvation which AAs are buried?
3
GenTHREADER
An Efficient and Reliable Protein Fold
Recognition Method for Genomic Sequences David
T. Jones (1999)
What a good presentation! B. Raveh (2003)
4
GenTHREADER overview
Query sequence MTYKLILNGKTKGETTTEAVD
AATAEKVFQYANDNGVDGEWTYTE
Templates

• For each template (in the Brookhaven PDB)
• Construct a profile sequence
• Align with query sequence
• Calculate structural parameters (to be
  continued)
• send parameters to a well-trained NEURON NETWORK
  (like PSIPred)
• OUTPUT match confidence alignment

5
STAGE 1 Building a profile for each template

• Start with sequence of template
  peptideMTPAVTTYKLVINGKTLKGETTTKAVDAETAEKAFKQYAN
  DNGVDGVWTYDDATKTFTVTC
• Run BLASTP on OWL non-redundant protein sequence
  data bank, with sequence as input.
• Take all sequences with E-Value lt 0.01.
• Align using MULTAL multiple sequence alignment
  method.
• Construct a sequence profile based on BLOSUM 50
  matrix.

6
STAGE 2 Align sequence with a profile
MTYKLILNGKTKGETTTEAVDAATAEKVFQYANDNGVDGEWTYTE
SCORE ?
Length of query sequence ?
Length of alignment itself ?
Length of template profile ?
7
STAGE 3 calculate (some) structural parameters
In stage 2, the sequence was aligned to a profile
of the structure.
The aligned sequence is now imposed on the 3D
structure of the template, and used for ENERGY
POTENTIALS calculation.
8
STAGE 3 structural parameters (cont.)
E-Pair (pair interaction potential)

• an energy potential for the probability of the
  interactions observed in this structure.
• Distance and sequence separation between certain
  atoms of two different amino-acids are measured
  (Cß Cß , Cß - N, Cß O, etc.)
• Statistics of known structures were gathered and
  weighted.
• The observed interactions are compared to the
  statistics
• An energy potential is calculated
• In essence the smaller E-Pair, the better.

aa 39
aa 157
9
STAGE 3 structural parameters (cont.)
E-Solv (solvation potential)

• Degree of burial (DOB) for an amino acid the
  number of other Cß atoms located within 10Å of
  the residues Cß atom
• In general, hydrophobic amino acids like to be
  buried, safely away from water.
• Hydrophilic acids might like the outside world
  better.
• Each amino acid DOB is calculated.
• Its compared to statistical occurrence.
• ?Esolv(AA,r) -RT ln( f(AA,r) / f(r) )

Cß
10Å
Cß
Cß
Cß
Cß
Cß
10
STAGE 4 send it all to the (trained) Neuron
Network
Ouput is a score between 0-1 translated to
confidence level (Low, Medium, High Certain)
11
See this page on the web
12
Who trains the Neural network?

• CAT numbers were used for comparing pairs.
• 9169 chain pairs
• 383 pairs shared a common domain fold ( should
  give a positive answer)
• The network was trained with these pairs.

13
Neural network black box?
14
Confidence assignment
CERTAIN
LOW
MEDIUM
HIGH
15
GenTHREADER what to do with it?

• Results on a classic test set of 68 proteins
• High true-positive rate 73.5 correctly
  recognized, 48.5 with CERTAIN.
• Extremely reliableEvery CERTAIN prediction
  was correct.
• Fast automatic method.
• For 22 of 68 proteins, alignment is over 50
  accurate.
• Lets go analyze the Mycoplasma Genitalium with
  it!

16
Whole Genome Analysis with GenTHREADER
Mycoplasme Genitalium genome analysis ONE DAY
ONLY!
17
ORF MG276 of mycoplasma gen. spotting a remote
homologue

• MG276 is an Adenine Phospho-ribosyl-transferase
  (but this information is not given to
  GenTHREADER)
• 1HGX is a template of other Phospho-ribosyl-transf
  erase.
• It has only 10 sequence identity with our MG276!
• It was found by GenTHREADER as a certain match
• E-Pair saved the situation!
• But how do we know its true?

18
Ligand binding site of 1HGX template
19
ORF MG276 of mycoplasma gen. supporting
evidence for 1HGX as a template

• We cheated all along

20
ORF MG353 of mycoplasma gen. an ORF with no
known function

• MG353 no homologues found in databases
• 1HUE is a template of an Histone-like protein
• Very low sequence similarity with our MG353.
• It was found by GenTHREADER as a certain match
• Striking similarity in DNA Binding regiondespite
  overall low sequence similarity

21
GenTHREADER improvements(McGuffin, Jones - may
2003)

• PSI-BLAST, PSI-PRED (2nd stuructures), some more
• Some Results

22
AB-INITIO FOLDING - ROSETTA (Simons et al 1997,
1999, Bystroff Baker 1998, Bonneau et al
2001) Prediction of a protein fold from scratch?
Method I physically simulate protein
folding Problem CPU time Practical for short
peptides
APKFFRGGNWKMNGKRSLGELIHTLGDAKLSADTEVVCGI
APSITEKVVFQETKAIADNKD WSKVEVHESRIYGGSVTNC
K ELASQHDVDGFLVGGASLKPVDGFLHALAEGLGVDINAKH
Method II check probability for all possible
conformations Problem infinite search
space Solution use mother nature decrease
search space
23
Decreasing the search space using elements from
short peptides

• Take fragments of short peptides (3 residues 9
  residues long).
• Join them together
• Keep the 2nd structures constant.
• Play with the angles of loop residues.
• RESULT 200,000 decoy structures

24
In addition - I-Sites prediction 13
local-structure 3D motifs with sequence profiles

• Strong independence of motifs (fold-initiation
  sites?)

• complements secondary structure

25
Find the correct fold for a given sequence (back
to threading)

• P(sequence structure)
• Solvation
• 2nd structure amino acid (proline in helix,
  etc.)
• Pair Interaction
• ISites prediction for this sequence(3D motifs)
  did not contribute to performance
• Etc.

• P(structure) sequence independant
• 2nd structure packing
• Strand hydrogen bonding
• Strand assembly in sheets
• Structure compactness
• Frequency of I-Sites 3D motifs
• Etc.

26
RESULTS in CASP 4 Bakers a winner
27
We're done!