A Model to Determine Molecular
Weights of Proteins from Gel
Electrophoresis
By
Jose Ceja
Kamyar Ghods
CSUN/JPL-PAIR 2001
Outline
• Getting the data (Standards)
• Choosing a Model
• Getting the data
(Unknowns)
• Applying the model
• Results and conclusions
Getting the Data
• Two Methods were
used:
• Adobe Photoshop
• Spot Viewer
Choosing a Model
• Cubic of form:
log(MW)=a+b(RM)^2+c(
RM)^3
• Cubic of form:
log(MW)=a+b(log(RM))^
2+c(log(RM))^3
• Quad Cross Validation of
form:
log(MW)=a+b(RM)+c(R
M)^2
• SLIC
R-squared values
R^2 cubic
R^2 logvslog R^2 quadratic SLIC R^2
0.998474201 0.996826989 0.998101144 0.819965331
0.998374109 0.996438939 0.997927349 0.783873282
0.997613398 0.995592952 0.998122027 0.841725705
0.999277693 0.998768666 0.961205471 0.858399478
0.999349397 0.998984631
0.99926587 0.898717842
0.999532322 0.998752274 0.999007192 0.897073329
0.999156965 0.998995702
0.99913475 0.783475311
0.999683346 0.999667791 0.952916318 0.844505686
0.999350391 0.999374277 0.993653215 0.827286213
0.999704933 0.999575949 0.994743706 0.858422394
Cross Validation
Cubic Cross Validation
Average
5.10
4.92
4.85
4.76
4.58
4.38
4.13
3.89
4.08
Bias
STD
0.20
0.14
0.14
0.07
0.07
0.11
0.20
0.27
0.27
0.004341
0.010027
0.005736
0.011781
0.006705
0.008265
0.014715
0.011892
0.013554
b/w
w/b
w/b
w/w
b/b
b/b
b/w
w/b
b/b
Quadratic Cross Validation
Average Bias
STD
5.07
0.23 0.004153
4.98
0.09 0.010712
4.90
0.09 0.007903
4.77
0.05 0.007556
4.57
0.08 0.007943
4.34
0.15 0.009269
4.10
0.23 0.013361
3.98
0.18 0.012924
4.13
0.32 0.017821
Applying Our Model
• Collected unknown
data using Photoshop
• Spot viewer not
designed for 1D gels
and not well
understood.
• Applied best cubic model to
each gel.
Applying Our Model
• Created an average of Jose's cubic Avg. cubic Komy's cubic
our two data sets
0.998474201 0.998479979 0.998237747
• Applied cubic model
0.998374109 0.998422789 0.998217619
to all
0.997613398 0.998058698 0.998215558
0.999277693 0.998947849 0.998394636
• Each standard had 3
0.999349397 0.999502212 0.999557248
cubic fits
0.999532322 0.999666967 0.999730577
• Used data that had the 0.999156965 0.999388038 0.999513854
best cubic fit for each 0.999683346 0.999498821 0.999189923
standard
0.999350391 0.999090399 0.998709349
0.999704933 0.999573455 0.999354771
Jose’s Unknown
• Frog skin Gels @ 7
and 12% for males and
females
• Within the same gel
different lanes had
different bands.
• Male and Female
frog’s skin do not have
the exact same
proteins
7% Male & Female frog skin
250000
Molecular Weight (D)
200000
150000
Male7.5%-L6
Male7.5%-L5
Female7.5%-6
100000
50000
0
0
0.2
0.4
0.6
Relative Mobility
0.8
1
1.2
Male and Female Frog Skin @ 12%
350000
300000
Molecular Weight (D)
250000
200000
"Male @ 12%"""
Female @ 12 %
150000
100000
50000
0
0
0.2
0.4
0.6
Relative Mobility
0.8
1
1.2
Komy’s Unknown
• Comparing 3 methods
• Overall the Manual
method found the
most proteins and the
Amylase method
found the least.
• The replicates of each
gel were pickkin up
more and different
proteins.
Molecualr Weights vs Relative Mobility
250000
200000
Molecular Weights (D)
Amylase
DTT
150000
100000
50000
0
0
0.1
0.2
0.3
0.4
0.5
0.6
Relative Mobility
0.7
0.8
0.9
1
Molecualr Weights vs Relative Mobility
350000
Molecular Weights (D)
300000
250000
Amylase
200000
DTT
150000
100000
50000
0
0
0.1
0.2
0.3
0.4
0.5
0.6
Relative Mobility
0.7
0.8
0.9
1
Conclusions & Future Work
• We both found that the higher
concentrations found more proteins.
• Photoshop is more reliable for dense 1D
gels.
• Out of the four models we tried, the cubic
model was the best one.
• Further study is needed to find a true
function relating RM to MW.
Aknowledgements
• We thank CSUN/JPL-PAIR program,
especially Dr. Carrol, Dr. Clevenson, Dr.
Shubin, V. Hutchins and J. Handy.
• And our fellow students
Residuals
Cubic Residuals
0.00
0.00
0.01
0.01
0.01
0.02
0.01
0.01
0.00
0.00
0.00
0.01
0.02
0.01
0.00
Quad Residuals
0.01
0.13
0.11
0.02
0.02
0.02
0.01
0.03
0.03
0.00
0.05
0.05
0.01
0.24
0.23
0.00
0.00
0.14
0.02
0.03
0.06
0.25
0.12
0.04
0.01
0.06
0.02
0.08
0.23