what to do next

calculate the angles between strands and within strands for barrel.
calculate the distance matrix to see if we can find a distance matrix similar to the one in the excel file.

I. strands at the edge of sheet
1. read the Richardson's paper.
2 . Statistical analysis of hydrophobic and hydrophylic residues in strands at the edge of a sheet
(strand with many hydrophilic residues is preferable at the edge)
3. Long strand usually not at the edge of sheet ?
... many other Mark's ideas

II. strands in a strandon
1. Check and compare the length of loops between strands within a strandon and between strandons.
2. check H-bonds between strands in strandons: any relations (loop, sequence in strands, and something which we don 't know else) with H-bonds between hydrophobic or hydrophilic a.a.

paper about edges

read this paper.
maybe we can predict if a strand is at the edge or not.
http://www.pnas.org/cgi/content/full/99/5/2754

Found a good website

CoC website tells you the conserved residues for protein representatives.
example for Immunoglobulin
example for barrel
paper

Questions about barrels

1. How to calculate axis of a barrel? (if we have the axis, maybe it will help us to calculate the bended strand)

2. Any other methods to calculate the bended angle which doesn't need pdbsum info?

• maybe I'll use this method to calculate the angle of the planes of ribbon in one strand, so that we can calculate the bended angle. (Because the method which calculates the angle between residues seems doesn't work)

3. why 1p3hA is n*=4, S*=8? it seems to me it's S should be more then 8.

Jmol instead of Chime

I might use Jmol instead of Chime,
because Dr. Kister's computer can't run Chime, but I think Jmol should be ok.
Use same command as Chime, so I only need to change a little bit of my program.

test here by choose "Pop-Up Examples - Select"

problems in barrel

• can't detect by only looking to one strand, can be detected by see two strands. The core seems to content four strands (1pdr)
• strand is too short in pdbsum, but since it's bended, it should be short in pdbsum. (1kef A)
• check "1nte A", how do you know which loop of strand 4 should be considered as strand?

Due to problem2 and problem3, check two strands or more didn't help. Because you don't even know where the strands start and end.

I think we should monitor the direction of the strands no matter it's sheets or loop, but I didn't find out a good way yet.

angle method +-2 instead of +-1

1kef A
false detection in +-1 strand1
so I came out method with +-2

and for regular one, +-2 detect correctly too. (ig3p)

1kef A

+-1
1 71.91
5 74.31
4 0.00
3 89.92
2 26.81

+-2
1 3.37
5 10.16
4 22.14
3 44.78
2 14.76

1g3p

+-1
1 89.26
2 49.14
3 90.24
4 159.00

+-2
1 81.58
2 26.23
3 98.49
4 137.03

how to calculate angle

seems moltalk can do the work but didn't work online.

sybyl command "measure angle M1(wwwi.ca)-M1(xxxj.ca)-M1(yyym.ca)-M1(zzzn.ca)" should do the work, but I don't have enought space on shrp to store the pdb files.

So, I think I'll calculate by this formula

ACADEMIC CALENDAR

http://shrp.umdnj.edu/current_students/Academic_Calendar0506.pdf

angle in barrel

PROTEIN FOLDS IN THE ALL-B AND ALL-A CLASSES
Annu. Rev. Biophys. Biomol. Struct. 1997. 26:597–627
P.611 Figure 7

THE CLASSIFICATION AND ORIGINS OF PROTEIN FOLDING PATTERNS
Annu. Rev. Biochem. 1990. 59:1007-39
P.19 Figure B

barrel

It seems there's three conserved residues in space, but no program in my hand can do the superposition, because they are not in order.

1sty
leu25
phe34
val74

1i8da
leu28
phe61
val42

1efca
phe332
leu362
val388

Use distance method to detect 8 residues

Tried on 1bww and 1kcl
pretty strait forward, just check contact of those four strands(including side chain),
mark contacts residues between non-H-bond strands,
pick two residues on each strand which 1) there's a residue between them. 2)from these two residues, you can contact with 3 other strands.
then, you'll have 8 residues.

by this way, we know 8 residues in hbalign_9_19.xls are wrong,
two residues of 1kcl on K strand are not Y and V, should be W and Y.

example process as flow

1bww A
3 4 8 9
3	8
23@	71
25#!	72
	73*%
	74*
	75*$
	77$
88$@	35%!
90#	37*@!
9	4
ANS:	23,25,35,37,75,77,88,90:A
1kcl A 582 686
34 35 36 37
34	36
591$!	634@
593#@%	635@
596%	636@%
598%	638!
	640!
650$!	604%
652$#%!	606!
653%
654%
37	35
ANS:	591,593,604,606,636,638,650,652:A

idea for H-Bond alignment

We have problem align H-Bonds networks because we don' t have a position on two different network that we know should be align together.

We can just use the interlock residues as the point.
Just alignment 8 residues together, then we know which network we should choose.

idea

Dr. Kister mentioned to me twice that his student can detect interlock residues by distance matrix.
Maybe we can use that tool to automatic classify the protein.
ex: number of interlock detected, related position of interlocks.

question: can we use distance matrix to detect residues works like interlock in red or green motifs?

try to use this info to see if we can predict strand by sequence.
tried NN a,a&5,4,3 etc. accuracy 66%

Dr. Kister might find the reason why the direction is wrong.

Dr. Kister figured this out on his way to the research club between second floor and first floor.
He told me that if we correct the direction, there will be an interlock,
in order to have an interlock, there must exist two pairs of hydrophobic residues connect with each other,
but when those residues already connect to it's consicutive strand, it's not possible to connect to the strand on the other strandon, that's why it must stay out of the interlock and be on the edge.

function to our website

maybe I should add a function to our website
align two or multiple protein domains in same motif and mark the strand number.

learn

papers about motifs (ex: greek key)http://arjournals.annualreviews.org/doi/pdf/10.1146/annurev.biophys.26.1.597?cookieSet=1

A Comprehensive Analysis of the Greek Key Motifs in Protein-Barrels and-Sandwiches

2G compares to 1FKey interactions in the immunoglobulin-like structure of apo-neocarzinostatin: Evidence from nuclear magnetic resonance relaxation data and molecular dynamics simulations

maybe I can use idea in this page and modify to my algorithm
PS: the whole idea of SPratt2 might be useful to me.

pretty interesting site, which, you can query by a sequence, and it will tell you which cluster this protein belongs to, including binding site, keywords, GO terms, Scop, Interpro, Taxonomy, etc.
http://www.protonet.cs.huji.ac.il/index_1_4.php

test http://bioinfo3d.cs.tau.ac.il/MultiProt/

about idea on 10/06/05

although the idea of 10/06 are not always right (ex:iF) , but the idea is, we should also use direction information in our algorithm if it's not too complicated.

1B should be 1A and another idea

when I'm checking this graph on this website

I'm amazed that the thread will never cross each other.
so, when I check 1B, it crossed, and I found 1B is wrong.

use the same method, we'll know 8A is not good, and we found 1df0A356-514 has a 17'

3A might be 1F

contact maps for 1bww(1A) and 1cd0 (3A)


when I'm considering use contact maps to do the alignment,
I saw 1bww and 1cd0 are pretty similar in contact maps.
I go back to check the strands in 1cd0, it does have strand 1, but we delete it.
by this way, 1bww and 1cd0 will in same SM.

It seems like contact maps has more sensitivity (although it will has lots of noise)

maybe we can see some pattern in contact maps, maybe some pattern in frequency domain of contact maps.

unthreading the protein will not tied

seems like unthreading the sandwich protein will not tied.
when ever it seems to tied, actually it's a interlock, just wait there, and unthread the other end, it will also come to the same interlock, and unlock the tied, so it never tied.

wrong order and hydrophobicity

checked by using Strap
no significant relation.
seems some are caused by hydrophobicity(ex:4L), some are not(ex:4Q,5K).

5J should't be in SM#5

1	(4	5)	15	7	13
(2	3)	6	14	(8	9	12	11	10)

This one should not in SM#5

I'm not sure if there's other motifs has this type of erroe.