Motif Peptide Analysis

2015-01-13

azim58 - Motif Peptide Analysis
There is code for handling glam2 here and other potentially useful Java code

When an antibody binds to random peptides on an array, some of these
random peptides may be more important or relevant to the original antigen
than others. There may be a way to analyze these random peptides to find
out which ones may be most important.

Map of Work 11-29-11

1st idea about strategy before 071012
Here's the basic strategy I attempt to use (071012) here.

Find motifs in the peptides.

o To do this, I will use glam2 to compare every possible
combination of two peptides in the list. If two peptides share a
motif with a score greater than the GLAM2 Significance Threshold,
then this motif will be added to the list of motifs.
- Maybe I'll call this: generateAllMotifs

Group similar motifs so that they just represent one motif group (or

it could just be called a motif)
o To do this, I will compute a similarity score between every pair
of motifs using glam2. For example, motif 2 will be selected and
compared against all of the other peptides in the list after it. If
the score is below some threshold, then the motifs will be grouped
together. When this motif 2 is included in the list, every motif will
then be compared to motif 2 to see if they also should be included in
this same group. Therefore, this will be a recursive function.

Perhaps find the most representative sequence in the motif group. I

might be able to do this by finding the similarity scores of every
motif to a particular motif. I would then average all of the scores for
that motif. The motif with the highest average score would be the
representative sequence for the motif group.

Find how many peptides belong to each motif group

o glam2scan will be used to assign peptides to a group.

Find the B cell epitope score for each peptide belonging to a motift

group, and then average these B cell epitope scores and assign this to
the motif group

Plot the motif groups based on the number of peptides that belong to

them and their B cell epitope score. Then see where the ELISA confirmed
peptides fall within this plot. Perhaps the best peptides belong to
motif groups with good B cell epitope scores and a lot of peptides with
this same motif

Do a short blast for all of the representative sequences of the motif

groups. Then see if there are any overlaps in the different blast
result lists. These overlap proteins may be proteins which the
antibodies binding to these random peptides originally bound to.

I will make use of three basic programs to accomplish this.

GEMODA (I may not use this program after all) (see Gemoda 1.0): this

program will list all possible motifs when given several sequences
o Note that gemoda only tells you whether the similarity "signal"
you are detecting is substantially different from the background
noise.
o see note from Mark 7-6-11

GLAM2: this program will try to find the best motif when given

several sequences

GLAM2Scan: this program will try to see if certain sequences contain

a given motif

Bepipred: this program can assign a B cell epitope score
BLAST: used to determine the similarity of a sequence to the database

of known protein sequences.

Java Classes
Most recent version of Java classes can be found here
"C:\kurt\storage\CIM Research Folder\DR\2012\10-2-12\code\mpa src 10-3-12.zip"
backup of classes on google
https://docs.google.com/file/d/0B55PHkbittziaWdCa25naFNOOEE/edit
https://docs.google.com/file/d/0B55PHkbittziWmxpX2lNa3Z1Qlk/edit

List of classes in this code



Bepipred_Handler.java


Blastp_Handler.java


Glam2Scan_Handler.java


Glam2_Handler.java


ItemData.java


LargeFastaBlastResultInfo.java


MotifGroup.java


MotifGrouper.java


MotifPeptideAnalysis.java


Motif_With_Peptide_List.java


UsefulTools.java

BLAST protein database stored here
S:\Research\Cancer_Eradication\Users\kwhittem\DR\2012\9-26-12_database\nr

===========================================================================
setting up computer to run program 9-23-12

===========================================================================

Note that in some early work looking at this, I found that there were
2,044 motif groups. This may not have been entirely accurate.

===========================================================================
see also: Amino Acid Substitution Matrix

Related papers 2-13-13