EPA-670/2-74-067
August 1974
Environmental Protection Technology Series
                                         OF THE  KINETICS
                                 OF VIRAL llCTtVAtlON
                                       National Environmental Research Center
                                         Office of Research and Development
                                        U.S. Environmental Protection Agency
                                                  Cincinnati, Ohio 45268

-------
                                  EPA-670/2-74-067
                                  August 1974
A MATHEMATICAL ANALYSIS OF THE KINETICS

         OF VIRAL INACTIVATION
                  BY

            Robert M. Clark
         Betty Lou Grupenhoff
            George C. Kent
   Water Supply Research Laboratory
      Program Element No- 1CB047
NATIONAL ENVIRONMENTAL RESEARCH CENTER
  OFFICE OF RESEARCH AND DEVELOPMENT
 U.S. ENVIRONMENTAL PROTECTION AGENCY
        CINCINNATI, OHIO  45268

-------
              REVIEW NOTICE





This report has been reviewed by the National



Environmental Research Center, Cincinnati,



and approved for publication.  Mention of



trade names or commercial products does not



constitute endorsement or recommendation



for use.
                   11

-------
                          FOREWORD
Man and his environment must be protected from the adverse
effects of pesticides, radiation, noise and other forms of
pollution, and the unwise management of solid waste.  Efforts
to protect the environment require a focus that recognizes
the interplay between the components of our physical envi-
ronment—air, water, and land.  The National Environmental
Research Centers provide this multidisciplinary focus through
programs engaged in

•  studies on the effects of environmental contaminants on
   man and the biosphere, and

•  a search for ways to prevent contamination and to recycle
   valuable resources.

This report describes a mathematical model which can be used
to characterize the response of viruses to a disinfecting
agent.  Not only is the model itself presented, but a tech-
nique is described which can be used to estimate the model's
parameters.  Both the model and the estimation technique are
being used to analyze experimental information resulting from
disinfection studies.
                              A. W. Breidenbach, Ph.D.
                              Director
                              National Environmental
                                Research Center, Cincinnati
                             111

-------
                      ACKNOWLEDGEMENTS
Dr. P. V. Scarpino, Professor of Environmental Engineering,
University of Cincinnati, provided guidance and assistance
throughout all phases of this work.

Mr. Arthur F. Hammonds of the Water Supply Research Laboratory,
NERC-Cincinnati, EPA, and Mr. Richard L. Manning, Office of
Water Programs, Office of Air and Water Programs, EPA, Wash-
ington, D. C. / assisted in the data processing work presented
in this paper.

Miss Jacqueline E. A. Kent assisted in the development of and
programming of several of the equations utilized in this analysis,

Ms. Catherine Hall, University of Cincinnati,  assisted in the
preparation of this manuscript.

Miss Gruppenhoff is employed as an engineer by General Electric
Company in Cincinnati; Mr. Kent is employed by the Office of
Water Programs, Office of Air and Water Programs, EPA, Wash-
ington, D. C.
                             IV

-------
A MATHEMATICAL ANALYSIS OF THE KINETICS OF VIRAL INACTIVATION

                        INTRODUCTION
Pathogenic enteric viruses transmitted via the water route
present a potential hazard to public health because of their
resistance to natural or artificial disinfection mechanisms.
More than 100 different strains of enteric viruses, causing
such diseases as poliomyelitis, meningitis, jaundice, and
gastroenteritis, are excreted in human feces.  The six major
groups of enteroviruses responsible for these diseases are
polioviruses, coxsackieviruses A and B, echoviruses, adeno-
viruses, infectious hepatitis, and reoviruses.  Since these
viruses are able to survive in sewage, natural waters, and
water supplies, they may pose a health threat, particularly
as wastewater reuse becomes more common.^' ^

Of constant concern to public health officials is the ability
of viruses to pass through water treatment plants.  The chlorine
levels must be adequate, not only for bacterial disinfection
but for viral inactivation as well.  As a result of the need
for constant concern over proper disinfection levels, much
research effort has been devoted to the s,tudy of the basic
disinfection mechanisms.

Chick was probably the first investigator to attempt to under-
stand the laws of disinfe'ction by applying the principles of
first-order kinetics to bacteria and spore inactivation.3  Only
the experiments with anthrax spores conformed to first-order
kinetics, whereas bacteria apparently followed another pattern
of inactivation.  Subsequent studies have obtained results
that confirmed first-order kinetic inactivation for bacteria.

Many research investigations have been directed toward the
study of the inactivation of viruses and enteric organisms.
As a result of these studies, the process of inactivation
has been found to be dependent on the time of contact between
the organisms and disinfecting agent, concentration of disin-
fecting agent, temperature, and pH.  In addition, viruses may
form clumps of varying sizes and may cause aberrations due
to their existence in inactivation systems.4  one approach
to studying the interaction of these various factors is to
develop a kinetic model that will systematically account for
them.  The development of such a model and its application
are discussed in this paper.

-------
                      MODEL DEVELOPMENT

One of the major features in this model is the consideration
of clumping or aggregation and its effect in explaining the
devitalization process and associated aberrations.  For pur-
poses of this model, it is assumed that the virions exist
either as individual particles in a suspension or as aggregates
or clumps made up of two or more particles.5  Each individual
particle or aggregate will form a plaque-forming unit  (PFU)
before the viral suspension is subjected to a disinfecting
agent.  It is impossible to determine whether a PFU represents
a single infective unit.  If the suspension contains single
particles as well as clumps of various sizes, the disinfection
process will continue until the last particle 'in the largest
clump is devitalized.  When the clump is completely devitalized,
a PFU is destroyed, but it is obvious that a distribution of
different size clumps will lead to a non-uniform destruction
of PFU's thereby causing some unusual shapes in the disinfec-
tion curve.

In this discussion, it will be assumed that this distribution
of infective units represents the state of the suspension.
The percentage of aggregates or clumps of all sizes which
have been disinfected at any time represents the Nth state;
the percentage of undisinfected single particles represents
the N-lst state, etc.  For illustrative purposes, let us
assume a suspension in which the maximum clump size is com-
posed of three viral particles and with clumps composed of
two particles as well as single particles.  Following our
convention, state 1 is the percentage of undisinfected aggre-
gates with three virions; state 2, the percentage of undisin-
fected aggregates with two virions; state 3, the percentage of
undisinfected single particles; and state 4, the total per-
centage of aggregates  (clumps of 1, 2, and 3 viral particles)
that have been devitalized at any point in time.  Obviously,
under the action of a disinfectant, assuming ideal conditions,
state 4 would increase as the process continues until state 4
would be 100 percent.

We can impose a frequency distribution on the various states
in effect, assigning a percentage of the total plaque-forming
capability to each state.  The initial condition of state 4(84)
must equal 0 percent at time equal to zero or before the disin-
fectant acts.  The percentage of undisinfected singles plus
the percentages of clumps with two particles plus the percentage
of clumps with three particles would equal 100 percent when
time equals zero.

Associated with each state is a decay rate, k^, that represents
the probability of interaction of the destructive agent with
the undisinfected singles or aggregate.  The process of devital-
ization is assumed to take place in the following manner:  The

-------
clumps of three virions are reduced to two  surviving virions,
and the clumps of two are reduced to one  surviving virion  all
the way along the chain of states until the clumps are no  longer
infective and are registered as a decrease  in total PFU.

The set of differential equations that describes the devitali-
zation process, where S^(i = 1  .  .  .4),  the percent of plaque-
forming capability at each state  is:

                      dsl
                      a-r = -

                      dS2
                          = k2S2 - k3S3
                      dt~ = k3S3

These are  a  set of  linear first-order differential equations.
The parameters kj_  (i =  1 .  .  .  4) represents the devitalization
rate with  k4 = 0, and s9 is the initial condition of  state  i
with sO =  0  at t =  0.   X

The solution to Equation 1  is as  follows :


             0         e"klt                 e"k2t
S4 - klk2k3Sl  [(-k!) (k3-ki) (k2-ki) +  (-k2) (k3-k2)
        (-k3)(k2-k3)(k!-k3)  T  (k3)(k2)
                                                      i
                                                      J

           32   (_k2) (k3-k2)    (-k3) (k2-k3)    (k-3) (k2)


           [e~k3t  -  1].                                    (2)
The general closed  form  solution  to  a  set of differential
equations as  illustrated by  Equation 1 is given  by  the  fol-
lowing : 5

-------
                     I   (Wi---1We''Cnt"s°'         I3.
                     =l  'VV 'Vl- V • '    k?
where k>j.  When j=k,  (k.-k, ) =  1.
                        1   K
When j "Dressed as percent survival, the equation  could  be
written as percent survival =  100  - S^, where  S^  is  the last
or final state to be considered.

Figure 1 illustrates schematically the change  taking place
during an experiment.  Devitalized virus  in  an aggregate are
represented by a broken circle.  In a devitalization chain,
the value for kj_, which indicates  the rate of  transition from
one state into the next, differs for each state.6 There are
also differences between chains.   For example,  k3 in the first
chain may be smaller than k3 in the second chain.  This might
be attributed to different geometric configurations  and re-
sulting interferences.  We will assume, however,  that k3 is  an
average reaction rate  for state 3  in all  of  the decay chains.

Equation 2 can be reformulated in  the following manner :

                                                            (4)


where,
                    (k3-k1)
                      -k k s°        k q°
                       KlK3bl        K3b2
                                     (k3-k2)
                      -k k q°        k q°
                        1 2bl        K2b2
                                            _ _
                3 ~  (k2-k3) (k1-k3)    (k2~k3)   b3


              C0 = Sl + S2 + S3                             (5)

We know that as t-*-°°, S.->-100 percent; therefore, Cn-*100 percent.

Equation 4 forms the basis for the mathematical model of the
kinetics of viral inactivation we wish to examine.  However,
to use this equation, we must be able to estimate its param-
eters.

-------
                                              •*-  ( II
                          o
                        83          84         85
                                       0
                                       4
Figure 1.   Schematic illustration of the physical change
           taking place in  a  suspension of virions under
           the influence of a devitalizing agent.
           Devitalized virus  in  a clump is represented by
           a broken circle.   The values for X indicate the
           rates of transition from one state into the
           next in a devitalization chain.  S? is the
           initial percent  of plaque-forming capability
           in state i.

-------
                  ESTIMATION OF PARAMETERS
If we were to rewrite Equation 4 in terms of percent  survival,
we would have the following:
          100% - S° = -Cje-!  - C^-     - Cge-             (6)


or

                  y - -C^l* - C2e-k2t  - C3e-k3t          (7)


where y = 100% - S,.

For simplicity/ we shall assume that our  observations  are
equidistant, as in Figure 2, and that the difference in  the
successive abscissa values is h.  With the use of our  three-
term example, we find the value of the ith ordinate at
to +  (i-l)h, where to is the value of yo  at to,  is then:

    y± = -C^ expt-k-^Q +  (i-l)h] - C2 exp[-k2tQ  +  (i-l)h]

         -C3 exp[-k3tQ +  (i-l)h]                            (8)

or, if we make the following substitutions:

                                   -kih
                                  e  x  = u-j^;

                                   -k-,h
                                  e  1  = u2;

                                   -kih
                                  e  J-  = u3;

                        -k-Lt^ +  (i-l)h] = f1?

                -C2 exp[-k2tQ +  (i-l)h] = f2;

                -Cx exp[-k3tQ +  (i-l)h] = f3?


then, for five equidistant measurements,  we have:

                        fi + f2 + f3 - y±


                  flul + f2U2 + f3U3

-------
100.0
                                       TIME
      Figure 2.  Equally distant values for percent survival versus time,

-------
                  flUl + f2U2 + f3U3 =
                  flul + f2U2 + f3U3 =
                  flUl + f2U2 + £3U3 =
In general, the set of equations for N observations would be
as shown below:

flul
flul
fl + f2
+ f2U2 +
+ f2u2 +
+ h
f3U3
f3u3
- yo
= yl
= Y2
which would necessarily be satisfied identically.  If the con-
stants U]_, u2, and U3 were known  (or ^reassigned) , Equations 10
would comprise N linear equations in the three unknowns f^, f2,
and f3, and be solved exactly if N=3 or approximately by
least squares if N>3.

However, if the u's are also to be determined, at least six
equations are needed, and a difficulty occurs because the
equations are non-linear in the u's.  This difficulty can be
minimized by the following method.

Let u, , Up, and u., be the roots of the algebraic equation:


                   3      2
                  u  - a,u  - a2u - a., = 0                (11)


so that the left-hand member of Equation 11 is identified with
the product (u-u^)(u-u2)(u-u3).  To determine the coefficients
a]_, a2, 33, we multiply the first line in Equation 10 by ao,
and the second line by a2, and the third line by a^, and the
fourth line by -1, and add the results.  If use is made of the
fact that each u satisfies Equation 11, the result is seen to
be of the form:


                 Y3 " aly2 ' a2yl " a3Y0 = °

-------
A set of N-4 additional equations of similar type is obtained
in the same way by starting instead successively with the second,
third . .  .  (N-3)th equations.  In this way, we find that
Equations  10 and 11 imply the N-3 linear equations:8


                       Y2al + yla2 + yOa3 = y3
yN-2al
                                   %-4a3 =
Since the ordinates y^ are known if N=6 , this set generally
can be solved directly for a^ , a2 , and 33, or it can be solved
approximately by the method of least squares if N>6.

In theory, after the a's are determined, the u's are found as
the roots of Equation 11 and may be real or complex.  Equation 10
then becomes linear and the f's can be determined from the first
n of these equations or preferably by applying a least-squares
technique applied  to the entire set.

We have examined the situations in which there are only three
terms to analyze in Equation 2.  However, most often the situ-
ation will occur when there are n terms in the equation to be
solved.  This would take the form as follows:
         Sn = C0
                                             (14)
Assuming that there  are N points equally spaced at t=0 , 1, 2,
3  ... N-l, and  following  the  logic described in this paper,
we get a set of equations similar to Equations 8:
                                            fn =
flul H
flul H
h f2U2 H
h f^ H
h f 3u3 + . ,
h f 3u2 + . .
' • + f nun = S
'• + fnun= S
Again, following the  logic described earlier, we have the  fol
lowing N-n linear equations where the columns of data are
labeled 1 through n+1.

-------
(1)
ynal H
(2)
h yn-la2 H
(3)
h yn-2a3 +
h yn-2a3 +
(n) (n+1)
• • ' + Y0an = yn
' ' ' + ylan = yn+l
     YN-2al + %-3a2 + YN-4a3 + ' ' '  + YN-n-lan = yN-l
(16)
After the a's have been determined by least squares, the values
for the c's can be found as roots of the, following equation:


         un - a^11"1 - a2un~2 - ... - an_2u - an = 0       (17)


And once the u's have been found, the f's can be found from
Equation 15.

The application of. this approach presumes that the number of
terms that make up the model is known.  Generally this number
is unknown, and a major part of the analysis becomes the esti-
mation of the optimum number of terms describing the disinfection
process.  Even if the number of terms is known, the solution to
Equation 17 is often complex because of estimation errors in
determining the coefficients.  To make this analysis usable,
we must be able to determine the number of terms (number of
states), that make up the inactivation process.  The following
section describes a technique for estimating the number of
components that "best" describe the inactivation process.
OPTIMAL NUMBER OF TERMS

To determine the proper number of terms that will describe
the inactivation process, we would formulate the set of linear
equations shown in Equation 16.  In this set, the column
labeled n+1 is the response or dependent variable, and the
columns 1 through n are the independent variables.  Using
step-wise regression, we regress the independent variables
(1 through n) against the n+lst or dependent variable.^  As
each variable is forced into the equation, a value for its
coefficient is calculated.  Each coefficient has an associated
sign.   When the signed coefficient is substituted into Equa-
tion 17, it is possible that an equation with alternating
signs may result; for example, Equation 17 might look as
follows:


        anun - a,un   + a0un   - ... + a  ~u - a  =0      (18)
         U      l        z              n-Z     n
                             10

-------
According to Des Cartes'  rule  of  signs:

          The number of positive  real  roots of  a
          real albegraic  equation either  is equal
          to the number Na of  sign changes in the
          sequence a0 , alf a2 ,  .  .  . an of co-
          efficients where vanishing terms are
          disregarded or  it  is  less than  Na by
          a positive even integer.

Since the decay coefficients in Equation  14 are the positive
real roots in Equation 18, we  can use  Des Cartes' rule to
give us an indication as  to  the number of terms which optimally
describes the inactivation process.  We will assume that when
the number of terms in the regression  equation  is one more
than the number of sign changes,  the optimal number of terms
has been identified, and  the variables in the regression equa-
tion are to be used in calculating kn.  The approach will be
discussed beginning with  the identification of the optimal
number of terms.

We can illustrate this approach by assuming a model of three
terms as follows:
      y = 20.00e-°-10t + 30.00e-°-30t + SO^Oe'0'         (19)
Table 1  (Page 12) contains values for Equation 19 which have
been generated at intervals of t=»0.50 to simulate a disinfection
curve.  Table 2 illustrates the way in which these data are
organized to solve for the coefficients in Equation 17.  As
shown in Equation 16, a matrix of data points is established
with n dependent variables.  In this case, 27 independent
variables have been constructed.  The value of yi - 100.00 is
the first value in the upper left-hand corner of the matrix,
and the value y2j = 5.7660 is the first value for the dependent
variable.  The second value for the first independent value is
y-L = 83.7858, and the second value for y28 = 5.4274.  This same
pattern is repeated throughout the matrix.

      Table 2.  MATRIX OF DATA FOR REGRESSION ANALYSIS
Var 1
100.00
83.786
Var 2
83.786
70.648
. . ., Var n ... Var 28
5.7660
5.4274
       2.0359    1.9338                           0.5203

       1.9338    1.8370                           0.4949
                             11

-------
Table 1.  VALUES FOR EQUATION 19 AT INTERVALS OF t=0.5
t
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
4.50
5.00
5.50
6.00
6.50
7.00
7.50
8.00
8.50
9.00
9.50
10.00
10.50
11.00
11.50
12.00
12.50
e-o. loot
20. 0000
19.0245
18.0967
17.2141
16. 3746
15.5760
14.8163
14.0937
13.4064
12. 7525
12.1306
11.5390
10.9762
10.4409
9.9317
9. 4473
8. 9865
8. 5483
8.1313
7.7348
7. 3575
6. 9987
6. 6574
6. 3327
6. 0238
5. 7301
e-0. 3001
30.0000
25.8212
22. 2245
19. 1288
16.4643
14.1710
12.1970
10.4981
9. 0358
7. 7772
6. 6939
5. 7615
4. 9589
4. 2682
3. 6736
3.1619
2.7215
2. 3424
2.0161
1.7353
1.4936
1. 2855
1.1064
0. 9523
0.8197
0. 7055
e-o. soot
SO. 0000
38. 9400
30. 3265
23.6183
18. 3939
14. 3252
11.1565
8. 6887
6. 7667
5. 2699
4. 1042
3. 1963
2. 4893
1.9387
1. 5098
I. 1758
0.9157
0.7132
0. 5554
0.4325
0. 3368
0. 2623
0. 2043
0.1591
0. 1239
0. 0965
y
100. 0000
83. 7858
70. 6478
59.9613
51.2329
44. 0722
38. 1699
33.2806
29. 2090
25. 7997
22.9287
20. 4969
18.4245
16. 6478
15.1152
13.7852
12. 6239
11.6039
10. 7030
9. 9027
9.1881
8. 5467
7. 9682
7.4443
6. 9675
6. 5321
t
13.00
13.50
14.00
14.50
15.00
15.50
16.00
16.50
17.00
17.50
18.00
18.50
19.00
19.50
20.00
20. 50
21.00
21.50
22.00
22.50
23.00
23.50
24.00
24.50
25.00
25.50
.-0. lOOt
5. 4506
5. 1848
4.9319
4. 6914
4. 4626
4.2449
4. 0379
3.8410
3. 6536
3. 4754
3. 3059
3. 1447
2.9913
2. 8454
2. 7067
2. 5747
2.4491
2. 3296
2.2160
2. 1079
2. 0051
1.9073
1.8143
1.7258
1.6417
1.5616
e-0. 300t
0. 6072
0. 5226
0. 4498
0. 3872
0. 3332
0. 2868
0. 2468
0.2125
0. 1829
0.1574
0.1354
0.1166
0.1003
0. 0863
0. 0743
0. 0640
0. 0550
0. 0474
0. 0408
0.0351
0. 0202
0. 0260
0. 0223
0.0192
0.0165
0.0142
-0. SOOt
0.0751
0.0585
0.0455
0.0355
0. 0276
0.0215
0.0167
0.0130
0.0101
0. 0079
0. 0061
0. 0048
0. 0037
o. 0029
0. 0022
0.0017
0.0013
0.0010
0. 0008
0. 0006
0.0005
0. 0003
0. 0003
0. 0002
0. 0001
0.0001
y
6.1330
S. 7660
5. 4274
5.1141
4. 8235
4. 5533
4. 3016
4. 0665
3. 8467
3. 6408
3. 4476
3. 2661
3. 0954
2. 9347
2.7833
2. 6404
2. 5055
2. 3781
2. 257?
2. 1437
2. 0359
1.9338
1.8370
1. 7453
1. 6584
1.5760
t
26.00
26.50
27.00
27.50
28.00
28.50
29.00
29.50
30.00
30.50
31.00
31.50
32. 00
32. 50
33. 00
33.50
34.00
34.50
35.00
35.50
36.00
36.50
37.00



-0. lOOt
1.4854
1.4130
1.3441
1.3785
1.2162
1. 1568
1. 1004
1.0467
0. 9957
0.9471
0. 9009
0. 8570
0.8152
0. 7754
0..7376
0.7016
0. 6674
0. 6349
0. 6039
0. 5744
0. 5464
0. 5198
0. 4944



e-0. 300t
0.0122
0.0105
0.0091
0. 0078
0. 0067
0. 0058
0. 0049
0. 0043
0. 0037
0.0031
0. 0027
0.0023
0. 0020
0.0017
0.0015
0.0012
O.OOU
0. 0009
0. 0008
0. 0007
0. 0006
0. 0005
0. 0004



e-o. soot
( : •!•>'
6. 
-------
Table 3 contains the results of the  application  of  the  step-
wise regression program to  the matrix of  data  in Table  2.
The equations resulting from each  step  are  as  follows:


                                        x28  - 0.9418X27  =


                            x28 - 1.1223X27  + 0.1402X23  =

               p Q           ^ *7           o "3            1
              x   - 1.1420x  + 0.1565X J - 0.000089x   =


   x28 + 0.0318X27 - 0.9398X23 + 0.07968x8  - 0.01124X1  =

Equation 22 combines the maximum number of  sign  changes with
the minimum .number of variables in the  equation  and is, there-
fore, selected as the equation governing  the number of  terms
in the disinfection equation.  This  matches identically with
the three terms used in the simulated data.  After the  best
estimate has been made of the number of terms  which makes up
the data, the next step in  the analysis is  to  estimate  the
decay coefficients in the equation.  This step is described
in the following section.

          Table 3.  RESULTS OF REGRESSION ANALYSIS
                        USING DATA FROM TABLE  2
0
0
0
0
(20)
(21)
(22)
(23)
Step
1
2

3


4



Var
27
23
27
1
23
27
1
8
23
27
Coefficient
0.94179753
-0.14022224
1.12226027
0.00008941
-0.15649791
1.14201651
0.01124259
-0.07967716
0.93985016
-0.03183525
ESTIMATION OF PARAMETERS

Decay Rates

Based on the data in Table 1, the parameters for Equation 19
can be estimated using the techniques outlined in Appendix A.
The first decay coefficient to be calculated will be that
associated with variable ,x23 and the calculation is as follows
                             13

-------
[(x24 - rlX23) - (x2 - r^1)]
          (23 _ 1}             =


24      23      28       27
24 - r,x2J)  - (x^b  - r,xz/)]
      L—r _ . _ ± _
        (23 - 27)

                                28   27   24   23   2
                                                            24
Substituting the average values for x   , x   , x   , x   ,  x  ,  and
x1, into Equation 24 yields the following values  for r:

                          r, = 0.85                        (25)

where
                               -k]h                        (26)
                         rl - e

From Equation 26, we can calculate k, as follows:

                                                           (27)
                    kx = -  [ln(0.85)]/0.50


                    kx = 0.32


The decay coefficient associated with variable x  is calculated
as follows:

                In(r2) = ln(x2) - ln(2xX - x2)             (28)


                Ind^) = -  0.24


Substituting into Equation  27 for k2 we get the following:

                      k2 =  (0.24)/0.50                     (29)


                      k2 =  0.48

                                                27
The decay coefficient associated with variable x    is calculated
as follows:

                    In(r3)  = In(x28/x27)                   (30)


                    In(r3)  = ln(0.95)


Substituting into Equation  27 for k3 we get the following:

                      k3 =  (0.05)/0.50

                      k3 =  0.10

                             14

-------
Coefficients
Once the decay rates in Equation  19 have been estimated, the
values for the coefficients are relatively easy to obtain.
Values for each exponential term  can be caluclated at the
appropriate time interval and these values regressed against
the values of y in Table 1.  Stepwise regression can then be
used to estimate the coefficients (Appendix D).


EXAMPLE INACTIVATION PROBLEM

To illustrate the utilization of  this technique, it will be
applied to experimental data collected from a series of
electromicroscopy investigations  conducted by Gordon Sharp
at the University of North Carolina.*-®  Sharp prepared
electron micrographs of dilute preparations of T7 virus
that had been subjected to a devitalizing agent.

Figure 3 shows the inactivation curve, and Table 4 contains
the distribution of T7 coliphage  particles resulting from
these experiments.  Column 1 of Table 4 lists the group size
of the aggregates, that is, the number of particles in each
clump of virus.  Column 2 lists the number of groups in the
suspension, and Column 3 lists the number of particles in
each group.  Column 4 lists the percent of plaque-forming
capability that each group represents in the suspension.
For example, there are 770 groups in the suspension, but
610/770 or 79.1 percent of them are groups of single viral
particles, and 116/770 or approximately 15.1 percent of them
are groups of two viral particles, etc.

                   Table 4.  T7 VIRUS DATA

      Group    Number of    Number of    Plaque-forming
      size      groups      particles    capability (%)

        1         610           610           79.22

        2         116           232           15.06

        3          24             72            3.12

        4          12             48            1.56

        5           6             30            0.78

        61             6            0.13

       18           1             18            0.13

      Total       770         1,016          100.00
                              15

-------
    INACTIVATION OF COLIPHAGE
        BY  ULTRAVIOLET  RAYS
    50    100
IRRADIATION
TIME  (SECONDS)  AT  33 jjW/cm'
    Figure 3.  Inactivation of coliphage
                   16

-------
Table 5 contains the data  from  Figure  3,  at  intervals  of
5 seconds, arranged in  10  columns  of data.   Table  6  contains
the coefficients associated with each  set of variables as
they enter the stepwise regression equation.   It is  obvious
from the alternating signs that six variables will describe
the inactivation process.  The  decay coefficients per  minute
calculated from the techniques  outlined in Appendix  A  are
as follows:
            /
                           r!  =  1-56                       (32)


                           r2  =  2.12


                           r3  =  2.32


                           r4  =  2.68


                           r5  =  2.83


                           rc  =  4.68
                           D

Each of these values represents a  k^ in Equation 14, and each
value of e^it can be generated at various intervals of t by
the program in Appendix B.  If  all six values  of e'^it repre-
sented by Equation 32 are  regressed against  the values for y
as obtained from the graph in Figure 3 then  Table 7 contains
the values for their coefficients.  Using the  program  in
Appendix C, the values  for each predicted Sj_  (percentage of
plaque-forming capability), can  be  calculated.  The predicted
and actual values are shown in  Table 8.

When the regression is  performed,  the values  shown in  Table 7
result.  At the fourth  step of  the regression, the corrected
R2 begins to decrease which is  an  indicator  that the regression
should be terminated at that  point, and step  3 is, therefore,
used as the last step in the  regression analysis.  Equation 5
and the program contained  in  Appendix C, where k^ = 1.56,
k2 = 2.83, and k3 = 4.68,  yields the following values  for Sj_:
83 = 73.97%, 82 = 15.51%,  and Si = 11.30%.   Physically, this
means that there are ,73.97% singles, 15.51%  doubles, and the
rest of the particles amount  to approximately  11.30%.  The
comparison between the  results  obtained from  the model and
the electron micrographs is shown  in Table 8.  The agreement
seems reasonable.
                              17

-------
00
                                       Table 5.  DATA FROM DISINFECTION CURVE
           Var 1     Var 2     Var 3     Var 4     Var 5     Var 6     Var 7     Var 8     Var 9     Var 10





         100.00000  79.00000  58.00000  45.00000  36.50000  28.50000  25.00000  19.80000  17.50000  15.00000



          79.00000  58.00000  45.00000  36.00000  28.50000  25.00000  19.80000  17.50000  15.00000  13.00000
         000.16500  00.16000  00.15500  00.15000  00.14000  00.13500  00.13000  00.12000  00.11000  00.10500



         000.16000  00.15500  00.15000  00.14000  00.13500  00.13000  00.12000  00.11000  00.10500  00.10000

-------
Table 6.  RESULTS OF REGRESSION ANALYSIS
             USING DATA FROM TABLE 5
Step .
1
2

3


4



5




6





7






Var
9
6
9
6
8
9
5
6
8
9
4
5
6
8
9
1
4
5
6
8
9
1
4
5
6
7
8
9
Coefficient
0.84449432
0.16137199
0.57866191
0.22635578
-0.22446799
0.73339338
-0.14587300
0.36096394
-0.35799450
0.95901767
0.12430537
-0.26913616
0.29413744
-0.36433149
1.01961917
-0.03634866
0.29008852
-0.30158431
0.22952950
-0.50623862
1.13438561
-0.03743081
0.32109643
-0.43943162
0.20853501
0.25010067
-0.67970749
1.22741919
                    19

-------
               Table 7.  RESULTS OF REGRESSION
                  USING DR. SHARP'S INACTIVATION DATA
Step
1
2

3


4



Var
2
2
6
1
2
6
1
2
4
6
Coefficient
96.77
89.08
6.64
58.89
4.12
37.77
61.68
-10.82
15.24
34.70
Corrected R2
as a percent t value
99.501 117.26
99.549 32.05
2.88
99.933 19.67
0.92
20.82
99.932 6.69
-0.23
0.32
3.55
         Table 8.  COMPARISON BETWEEN MODEL OUTPUT
                      AND ELECTRON MICROGRAPHS
        Group size
   Percent
plaque-forming
  capability
  (counted)
   Percent
plaque-forming
  capability
 (predicted)


3
1
2
+4+5+6+18
79.22
15.06
5.82
73.97
15.51
11.30
                   SUMMARY AND CONCLUSIONS

The kinetics of viral inactivation have been examined from a
rational point of view.  A mathematical model based on the
radionuclide chain decay concept was formulated and a solution
technique developed that allows for estimations of the optimal
number of terms in the equation and for estimating the equation's
parameters.  With the use of data derived from electron micros-
copy, the model was tested and achieved reasonable results.
                             20

-------
Based on this discussion, it is obvious that the postulated
mathematical model and its solution techniques are superior
to others that have been formulated.  The approach outlined
in this report not only determines the number of aggregate
groups in the suspension, but the values for decay coefficients
as well.  There are some deficiencies in this approach, however,
and it is important that these be considered.  The approach
suggested here is statistical in nature and is, therefore,
subject to experimental error in the various estimations made.
More importantly, the estimates of aggregate size and concen-
tration are blind.  That is, if this approach estimates three
terms as optimal, there is no way to provide information on
the make-up of these aggregate groupings.  They might be clumps
of single, double, and triple particles, or clumps of 20, 21,
and 25 particles.  The values for the decay coefficients may
give some insight as to clump size, but these insights are
hardly sufficient.  This technique must be coupled with a
physical assay approach incorporating electron microscopy.
A research project that combines the elements of mathematical
analysis with electron microscopy is currently underway.
                              21

-------
                         APPENDIX A
In this appendix, the mathematical justification for the
techniques used in the section entitled "Estimation of
Parameters" is developed.  Table 9 contains the first three
values for the individual terms which make up the variables
1, 2, 23, 24, 27, and 28, as shown in Table 2.  The first
variable to enter the stepwise regression equation is x27,
as shown in Equation 20.  Looking at variables 27 and 28 in
Table 9, it is obvious that the term labeled f$ dominates
variable 27 and is most highly correlated with variable 28,
while the terms f^ and f.2 in variable 27 are relatively
insignificant.  The next variable to enter the stepwise
regression equation is variable 23, and it can be seen that
terms ±2 and f$ in variable 23 are significant but that term
f^ is insignificant, and finally in variable 1, terms fi, f2/
and f3, are all significant.  It can be concluded from this
that a variable enters the regression equation when one of
the terms which comprise it is significant enough to alter
the rate of change of the entering variable.  Therefore, we
would expect that variables would enter the regression equation
with alternating signs associated with their coefficients,
since the entrance of each variable into the equation signifies
a significant change in the functions slope.  Moreover, we
would expect that the variables entering the equation with
alternating signs represents the maximum change in the slope
of the function with respect to the other variables in the
regression equation.  Using Equation 19 as an example, we
would, therefore, attempt to find a u^ such that bn in the
following equation is a maximum relative to its adjacent
variables:

fl(uf1 - u.uj) + *,(„»« - »±u» + f3(uf 1 - u.u") = bn    133)


or in a more simple form, we would attempt to find a u-^ such
that b23 is a maximum relative to bi and b27 in the following
set of equations:

                        x2 - U..X1 = b1                     C34)


                      x24 - u.x23 = b23


                      x28 - u.x27 = b2?
                              22

-------
Table 9.  FIRST THREE VALUES FOR SELECTED VARIABLES
Var
1
2
23
24
27
28
Time
0
0.5
1.0
0.5
1.0
1.5
11.0
11.5
12.0
11.5
12.0
12.5
13.0
13.5
14.0
13.5
14.0
14.5
Cie-0.500t.
50.0000
38.9400
30.3265
38.9400
30.3265
23.6183
0.2043
0.1591
0.1239
0.1591
0.1239
0.0965
0.0751
0.0585
0.0455
0.0585
0.0455
0.0355
C26-0.300tt
30.0000
25.8212
22.2245
25.8212
22.2245
19.1288
1.1064
0.9523
0.8197
0.9523
0.8197
0.7055
0.6072
0.5226
0.4498
0.5226
0.4498
0.3872
C3e-»-"°^
20.0000
19.0245
18.0967
19.0245
18.0967
17.2141
6.6574
6.3327
6.0238
6.3327
6.0238
5.7301
5.4506
5.1848
4.9139
5.1848
4.9139
4.6914
y
100.0000
83.7858
70.6478
83.7858
70.6478
59.9613
7.9682
7.4443
6.9675
7.4443
6.9675
6.5321
6.1330
5.7660
5.4274
5.7660
5.4274
5.1141
**!•
tf2-
                         23

-------
Therefore, u^ can be calculated from the  following  equation:
(x2  -
            -  (x
                24
23
f
(x
24
23,    , 28
  )  - (x
                          22.
          (1 - 23)
                                        (23 - 27)
                                                             (35)
                                                     23
 since we  know that  bn is  a maximum round the point x
 Table 10  confirms that this is in fact the case for variable 23.
            Table 10.   bn FOR SELECTED VALUES OF Uj_
x2 - u x1
«"• LI n.Ch
-2.2800
-1.4613
-0.7931
x24
0
0
0

.5839
.5583
.5000
28
x
0.
0.
0.
27
- U2x
4857
4449
4569
                1       27
 For  variables  x  and x  ,  this  computation is impossible since
 there  are  no variables which can be  used to make a computation
 similar  to Equation 2.

 However, for the  decay coefficient associated with x^-, the
 following  equation can be  developed  using the properties of
 infinite series:
           un(x2 - u
                            un+1(x1  -  x2)  as  n
                               (36)
 Therefore,
                      = ln(x2) - ln(2x1 - x2)
                                                             (37)
                                            27
For the decay  coefficient  associated  with x  ,  the following
relationship can be  developed:
                 n  28     n+1  27    n
                u.x   ~ui  x   -*  0  as  n -> «>
                                                             (38)
Therefore,
                  ln(u±) =  ln(x28)  -  ln(x27)
                                                             (39)
                              24

-------
                 APPENDIX  B

PROGRAM  FOR  GENERATING  VALUES OP  e~kit

          AT GIVEN  INTERVALS  OF t
     DlMtNSION A(15),b< 15), C( 15)
     MR = 1
     MU = 15
   5 READ(MR, 10 ) T , • M
   10 FORMATIF5.2, 12)
     IF IN)70, 70,15
   15 REAO(MR,20) (A( J), J = 1,N)
     READ(MR,20) (C( J) ,J=1,N)
   20 FGRMATI 10F8.0)
     E=2. 71828
     S = 0.0
     Y=0.0
     WRITl(MW,25)
   25 FORMAK • l'//26X, '-XT' )
     V»RITE(M^,30)
   30 FORMAT(25X, 'E    TABLE')
   35 FORMAT ( 25X, ' ---------- ')
     VvRITE(MW,40) (A( J) , J=1,N)
   40 FORMAT«//10X,  3(3X, '-« ,F5.3, 'T1 ) )
     fcRITE(MW,45)
   A5 FORMAT(5X,'T',6X,'E«,  2 ( 9X , • E • ) , 1 2X , • Y • )
   t>0 DO 60 I = 1,N
   t>5 B( I )=C( I )*E»»(-A( I )»S)
   60 Y = Y4-B(I)
     WRITEIMW.65)  S , ( B ( K ) ,K= 1 , N ) , Y
   05 FORMAT(/3X,F5.2,2X,11F10.4)
     Y = 0.0
     IF  (fid)-. 0001)     5,50,50
   10 STOP
     EMD
                       25

-------
                             APPENDIX  C

                 PROGRAM  FOR CALCULATION
    DIMENSION A(16),B(16),SUO),TEMP<10),STOR                                       009
    STOR(1)=STOR(1)«A(J)                                              010
 30  CONTINUE                                                          Oil
    U1=TEMP(1)/STOR(1)                                                012
    IF( 1-1)45,35,45                                                   013
 35  S(1)=B(1)«U1                                                      014
    WRITE  (3,40)  S(l)                                                 015
 40  FORMAT  (•!'////'  S(l>  =  «,F12.2///>
    GO  TO  41                                                          017
 45  TEMPI2I-1.0                                                       017A
    STCR(2)=1.0                                                       01TB
    DO  65  J = 1,N                                                       018
    IFU-J)SO,65,50                                                   019
 iO  STOR(2)=STOR(2)»A(J)                                              020
    TEMP(2) = TEMP(2)«(A(JI-AI I ) )                                       021'
 65  CONTINUE                                                          022
    U2=STOR(2)»SU)/TEMP(2)                                           023
    IF(I-2)80,70t80                                                   024
 ?0  S(2)=(B(2)-U2)«Ul                                                 025
    WRITE  (3,75)  S(2)                                                 026
 75  FORMAT  ('  S(2) =  SF12.2///)                                      027
    GO  TO  41                                                          02B
 00  TEMP(3)=1.0                                                       028A
    STOR(3)=1.00                                                      028B
    DO  100  J=2,N                                                      029
    IF(I-JI85,100,85                                                  030
 85  STOR(3)=STCR(3)»A(J)                                              031
    TEMP(3)=TEMP(3)*(A(JI-AII»                                       032
100  CONTINUE                                                          033
    U3=STOR(3)»S(2)/TEKP(3)                                           034
    IF(1-3)115,105,115                                               035
105  S(3)=(B(3)-U2-U3)»U1     v                                        036
    WRITE  (3,110)  S(3)                                                037
110  FORMAT  ('  SO) =  ',F12.2///J                                      038
    GO  TO  41                                                          039
115  TEMP(4) = UO                                                       039A
    STOR(4)=1.0 '                                       ,              039B
    DO  135  J=3,N                                                      040
    IF( I-JU20, 135,120                                               041
120  STCR(4)»STCR(4)»A(J)                                              042
    TeMP(4)=TffMP(4)*(A( J)-A( I I )                                       043
135  CONTINUE                                                          044
    U4=STOR(4)»S(3)/TEMP(4)                                           045
    IK 1-4)150,140,150                                               046
140  S(4)=(B(4)-U2-U3-U4)«U1                                           047
    WRITE  (3,145)  S(4)                                                048
145  FORMAT  ('  S(4) =  SF12.2///)                                      049
    GO  TO  41                                                          050
150  TfMP(5)=1.0                                                       050A
    STCR(5)=1.0                                                       050B
                                    26

-------
    DO 170 J = 4,N
    IHI-J>155,170,155                                                052
155 STOR(5)=STOR(5)»A(J)                                              ot,3
    TEMP<5»=TEMP(5)«(A(J)-AU))                                       054
170 CONTINUE                                                          Ot,5
    U5=STOR(5)«S(4)/TEMP<5)                                           056
    IF(1-5)185,175,185                                                057
175 S(5)=(B(5>-U2-U3-U4-U5)«U1                                        058
    WHITE (3,180) SI5)                                                059
180 FORMAT  (• S(5) =  «,F12.2///)                                      060
    GO TO 41                                                          061
185 TEMP(6)=i.O                                                       061A
    STOft(6)=1.0                                                       061B
    00 205 J=5,N                                                      062
    IF(I-J)190,205,190                                                063
190 STOK(6)=STOR<6J»A(J)                                              064
    TFMP(6> = TeMP(6)«U( J)-A< I »                                       065
205 CONTINUE                                                          066
    U6=STOR(6)«S(5)/TFMP(6I                                           067
    IF(I-6)220,210,220                                                068
210 S(6)=(8I6)-U2-U3-U4-U5-U6)»U1                                     069
    WRITE 13,215) S(6)                                                070
215 FLRWAT  I' S(6) =  ',F12.2///)                                      071
    GO TO 41                                                          0/2
220 TEMP(7)=1.0                                                       072A
  . STOR(7)=1.0                                                       072B
    DO 240  J=6,N                                                      073
    IF(I-J)225,240,225                                                074
225 STOR(7)=STOR(7)«A(J)                                              075
    TEMP(7)=TEfP(7)*(A(J)-A(I))                                       076
240  CONTINUE                                                         077
    U7=STOR(7)«S(6)/TEMP(7)                                           078
    IF( 1-7)255,245,255                                                079
245 S(7)=(B(7)-U2-U3-U4-U5-U6-U7)»U1                                  000
    HR1TE (3,250) 5(7)                                                081
250 FORMAT  (• S<7) =  SF12.2///)                                      082
    GO TO 41                                                          083
255 TEMP(8)=1.0                                                       083A
    STOR(8)=1.0                                                       083B
    00 275  J=7,N                                                      084
    IF(I-J)260,275,260                                                085
260 STCR(8)aSTOR<6)»A(J)                                              086
    TEMP(8)=TEMP(8)»(A( Jl-AI I ) )                                       087
275 CONTINUE                                                          088
    U8=STOR(8)«S(7)/TEMP(8)                                           089
    IF(I-8)290,280,290                                                090
280 S(8)=(B(8)-U2-U3-U4-U5-U6-U7-U8)«Ul                               091
    WRITE (3,285) S(8)                                                092
285 FORMAT  (' S(8) =  «,F12.2///)                                      093
    GO TO 41                                                          oq«
290 TEMP(9)=1.0                                                       094A
    STOR(9)=1.0                                                       094B
    DO 300  J=8,N                                                      °95
    IKI-J)295,300,295                                                °96
295 STOR(9)=STOR(9)»A(J)                                              °97
    TEMP(9)=TEMP(9)«(A(J)-A(I))                                       098
300 CONTINUE                                                          °"
    U9=STOR(9)«S(8)/TEMP(9)                                           100
    IF(I-9)315,305,315                                                l°l
305 S(9)=(B(9)-U2-U3-U4-U5-U6-U7-U8-U9)«U1                            102
    WRITE (3,310) S(9)                                                l03
                                    27

-------
'HO FORMAT (' S(9) = ',F12.2///)                                       104
    GC TO 41                                                           105
315 TEMP(10>=1.0                                                       105A
    STOR(10)*1.0                                                       10!>8
     DO 325 J=9,N                                                      106
    IFU-J)320,325,320                                                 107
320 STORI10)=STOR(10)»A(J)                                             100
    TfcMP(10)=TEMP«lO)*(A(J)-AU))                                      1°9
325 CONTINUE                                                           HO
    U10 = STOR(10)»S(9)/TEMP(10)                                         HI
    IF( 1-10)41,330,41                                                  112
330 S(10)»(B(10)-U2-U3-U4-U5-U6-U7-U8-U9-UIO)»UI                       113
    hRITE (3,335) S(IO)                                                114
335 FORMAT (' S(10) =  ',F12.2///1                                      US
 41 CONTINUE                                                           116
    GO TO 2                                                            H6A
 42 STOP                                                               117
    END                                                                118
                                      28

-------
                                APPENDIX D

                     STEPWISE  REGRESSION  PROGRAM
c     1130 STEPWISE MULTIPLE DEGRESSION PROGRAM,  3/14/66                    0010
C     PHASES 1 AND 2 CAN. BE OVERLAID TO CONStRVE  CORE.  THE STtPS TO         0020
C     READY PHASES \ AND 2 FOR OVERLAY ARE                                 0030
C        1. SET UP A COMMON AKEA CONSISTING OF  R U , X6AR, S IGMA.F IN,          0040
C        FQUT, DBS, NVAR, NOBS, NINDV, IRES, IFA.                                0050
C        2. SET SIGMA ANO DATA EQUIVALENT IN PHASE  2.                       0060
C        3. REPEAT PHASE 1 DEFINE FILE STATEMENT  IN  PHASE 2,                0070
C        4. REMOVE STATEMENT 101-3 FROM PHASE 1 AND  INSERT IT               0080
C        BEHIND DIMENSION COMMENTS CARD IN PHASE  2.                         0090
C     PHASE 1. TRANSFORM ORIGINAL DATA, COMPUTE AND  PRINT MEANSt            0100
C     STANDARD OEV I AT lOiSiS , ANQ SIMPLE CORRELATION CUEFFIC IENTS.             0110
C     DIMENSIONS                                                           0120
      IMPLICIT REAL*8(Ai-H,C-Z)
      DIMENSION DATA(30),CONST(12), ITRANI 30 I , JTRAN< 30 ) ,KTRAN< 30 > ,LTRAN( 3    0130
     10)                                                                   0140
      DIMENSION RIJ(30,30),XBAR(30>, SIGMA I 30 ) , AIO( 18 )                      0150
      DIMENSION SIGB(30) ,8(30) , 10(30)                                      0160
C     EQUIVALENCES                                                         0170
      EQUIVALENCE < S IGMAC 1 ) .DATA (1) )                                       0180
C     DEFINE DATA FILE                                                     0190
      DEFINE, FILE 10 ( 1000, 60, U, IFA )
C     STATEMENT LABEL  101 IS NOT REFERENCED. IT MARKS  THE FIRST             0210
C     EXECUTABLE STATEMENT OF THE SOURCE PROGRAM.                          0220
C     ICCM  IS FIXED DECIMAL REPRESENTATION OF ALPHABETIC  COMKA.             0230
  101 ICCM=27456                                                           0240
C     INITIALIZE DATA FILE                                                 0250
      IFA=l                                                                0260
C     READ  1.0.                                                            °270
      READ (U It ENO=999) ( AID ( I ) , 1*1 , 18 )
    1 FORMAT! 18A4)
C     REAP CONTROL CARD                                                    0300
      REAOU,2)NVIN,NVAR,NOBS,NTRAN,NCCNS,FIN,FOUT,IRES                    0310
    2 FORMAT(2I2, 14, 2 12* 2F6. 3, 1 1 )
      IF(FIN-FOUT)1020,690,690                                              331
  690 IF(NTRAN)100Q|730,700                                                0340
C   '  READ TRANSFORMATION CARDS                                            035°
  700 READ(l,71HITRANm,,JTRANm,KTRANm,LTRANm,l = l,NTRAN>             0360
   71 FQRMAT(36I2)                                                         037°
      IF(NCONS) 1000, 730, 720                                                038°
C     READ CONSTANT CARD                                                   °39?
  720 READ(1,72MCCNST
-------
C     X(JI=X(K)                                                              0610
  760 DATA(JJ)=DATA(KK)                                                      0620
      GO TO 850                                                              0630
C     X(J)=-X(K)                                                             0640
  HO DATA!JJ)=-OATA!KK)                                                     0650
      GO TO 850                                                              0660
C     X(J)=LOG X(K)                                                          0670
  780 DATA(JJ)=DLOG(DATA(KK))                                                0680
      GO TO 850                                                              0690
C     X(J)=1/X(K)    •                                                        0700
  790 DATA(JJ)=1.0/CATA(KK)                                                  0710
      GO TO 850                                                              0720
C     X( J)=X!K)+X!L)                                                         0730
  800 DATA!JJ)=OATA(KK)+DATA(LL)                                             0740
      GC TO 850                                                              0750
C     X(JI=X(K)*X(l)                                                         0760
  810 DATA!JJ)=DATA(KK)»DATA(LU                                             0770
      GO TO 850                                                              0780
C     X(JJ=X(K)/X(L)                                                         0790
  820 DATA!JJ)=DATA(KK)/DATA!LL)                                             0800
      GO TO 850                                                              0810
C     XI J)=X(K)+C                                                         0820
  (330 DATA( JJ )=DATA ! KK)+CCNST
-------
                                        ...M,

C      feRF°RK STEPWISE CALCULATIONS AND PRINT RESULTS.             1230
C     INITIALIZE
      DO 190  I=1,NVAR
      SIGB(I)=0.0
  190 Bii)=o.o
      NENT=0
      OF-OBS-l.O
      NSTEP=-i
C     TRANSFORM SIGMA  VECTOR  FROM  STANDARD DEVIATIONS TO SQUARE             1320
C     ROOTS OF SUMS CF  SQUARES.                                             1330
      DC 310  I=1,NVAR                                                       1340
  310 SIGMAl I > = SIGMA( I ) « ( CBS-1 .0 ) •«. 5                                       1350
C     BEGIN STEP NUMBER NSTEP.                                              1360
  i!00 NSTEP=NSTEP+1                                                         1370
      STDEE=( 
-------
      RSQP = RSC » 100.
      WRITE(3,59)  RSCP
   59 FORMAT!'  PERCENT VARIATION EXPLAINED R-SQ = ' , F 15. 3 )                     1773
      CRSQ = !.-(( l.-RSQ)«(CBS-l.))/(OBS-DEPV-1.)
      CRSGP = CRSG » 100.
      WRITE(3,84)  CRSGP
   84 FORMAT!1  CORRECTED R-SG AS A  PERCENT*',F20.3)
      IDFN=OBS-DF-2.0                                                        1800
      IDFD=DF+1.0                                                             1810
      F=(SIGMA(NVAR)«*2-(STDEE»*2)*!DF+1.0))/((OBS-DF-2.0)*STOEE«»2)         1820
      WRITE(3,66)IDFN,IDFD,F                                                 1830
   66 FORMAT!'  GOODNESS OF FIT OR OVERALL F,F(',I 3,',',I 3,') = ',F8.3)         1840
      WRITE(3,60)BSUEC                                                       1850
   60 FORMAT!*  CONSTANT TFRM=•,I8X,F16.8)                                    1-860
      WRITE(3,61)                                                             1870
   61 FORMAT!'OVAR        COEFF              STD OEV            T VALUE'     1880
     1)                                                                       1881
      WRITE<3,62>                                                             1890
   62 FORMAT!•                                 COEFF*)                        1900
      DO 430 1=1,NIN                                                         1910
      J=IO(I)                                                                1920
      T = L« D/SIGB! I )                                                         1930
      WRITE(3,63)ID( I ),B(I),SIGB(I),T                                        1940
   63 FORMAT!'  • , I 3 , F18.8,F20.8,F 18.8)                                       1950
  430 CONTINUE
C     COMPUTE F LEVEL FOR MINIMUM VARIANCE CONTRIBUTION VARIABLE             1960
C     IN REGRESSION EQUATION.                                                1970
      FLEVL = VMIN«DF/RU(NVAR,NVAR)                                            I960
      IF(FOUT + t-LEVL)460,460,450                                              1990
C     INITIALIZE FOR REMOVAL OF VARIABLE K FROM EQUATION.                    2COO
  4-iO K=NMIN                                                                 2010
      NENT=0                                                                 2020
      DF=OF+2.0                                                              2030
      GO TO 500                                                              2040
C     COMPUTE F LEVEL FOR MAXIMUM VARIANCE CONTRIBUTION VARIABLE             2050
C     NOT IN EQUATION.                                                       2060
  460 FLEVL=VMAX«DF/(RIJ(NVAR,NVAR)-VMAX)                                    2070
      IF(FLEVL-FIN)600,600,470                                               2080
C     INITIALIZE FOR ENTRY OF VARIABLE K INTO EQUATION.                      2090
  470 K=NMAX                                                                 2100
      NENT=K                                                                 2110
      GO TO 500                                                              2120
C     OUTPUT FOR VARIABLE DELETED                                            2130
  480 WRITE(3,64)NSTEP,K                                                     2140
   64 FORMAT!'OSTEP NUMBER ', 12 , 10X,'DELETE VARIABLE ',12)                   2150
      GO TO 425                                                              2160
C     UPDATE MATRIX                                                          2170
  bOO DC 540 I=1,NVAR                                                        2180
      IF! I-K)510,540,510                                                     2190
  510 DO 530 J=1,NVAR                                         •               2200
      IF(J-K)520,530,520                                                     2210
  520 RIJ(I,J)«RIJ(I,J)-RIJ(I,K)*RIJ(K,J)/RIJ                                            2310
                                      32

-------
5BO CONTINUE
    RIJ(K,K)=1.0/RIJIK,K)
    GO TO 200
600 IF!IRES)610,640,610
    PRINT RESIDUALS
blO IFA=1
    lriRITE(3,67)
 67 FORMAT!'0 OBS        ACTUAL
    WRITE(3,69)
 69 FORMAT! •
    DO 630  K=1,NG6S
    READ!10'I FA){DATA!I),I = 1,NVAR)
    EST=BSUBO
    DO '620  1 = 1,NIN
    J=ID(I)
620 EST = EST + B( I)»DATA(J)
    RESID  = DATAdMVAR)-EST
    XNORD  = RESID/STDEE
    IF»DA6SlXNORO)-3.191,92,92
ESTIMATE
RESIDUAL
NORMAL')
                          DEVIATE')
91
92
30
94
31
93
68
630
C
640
999
C
C
1000
C
C
C
1010
C
C
1020

IF! DABS (XNORD 1-2. 193,94,94
WRITE(3,30)K,DATA(NVAR) , EST , RES ID, XNORD
FORMAT!' ' ,I4,4F12.2,« ««')
GO TO 630
WRITE!3,31)K,DATA(NVAR),EST,RESID,XNORD
FORMAT!' ' ,I4,4F12.2, ' »')
GO TO 630
WRITE(3,68)K,DATA«NVAR) , EST , RES ID, XNORD
FORMAT!' »,I4,4F12.2)
CONTINUE
NORMAL END OF JOB
GO TO 101
CALL EXIT
ERROR. NIN, NENT, VMIN, NCONS, OR NTRANS IS NEGATI
FOR CONTROL CARD ERROR.
STOP1
ERROR DEGREES CF FREEDOM =0. EITHER ADD MORE DATA


VE. CHECK


OBSERVATIONS





OR
DELETE ONE OR MCRE INDEPENDENT VARIABLES. SAMPLE SIZE MUST EXCEED
NUMBER OF INDEPENDENT VARIABLES BY AT LEAST 2.
STOP2
ERROR. F LEVEL FOR INCOMING VARIABLE IS LESS THAN
OUTGOING VARIABLE.
STOP4
END


F LEVEL FCR









2320
2330
2340
2350
2360
2370
2380
2390
2391
2392
2400
2410
2420
2430
2440
2450
2460
2461
2470
2471
2480
2481
2482
2483
2484
2485
2486
2487
2490
2500
2501

2520
2530

2550
2560
2570
2580
2590
2600
2610
   33

-------
                          REFERENCES


 1.  Berg, G., "Virus Transmission by the Water Vehicle,"
     I. Viruses, Health Lab Science, 3:86 (1966).

 2.  Berg, G., "Virus Transmission by the Water Vehicle,"
     III. Removal of Viruses by Water Treatment Procedures,
     Health Lab Science, 3:170 (1966).

 3.  Chick, H., "An Investigation of the Laws of Disinfection,"
     Journal of Hygiene, 8:92 (1908).

 4.  Berg, G.; Clark, R. M.; Berman, D.; and Chang, S. L. ,
     "Aberrations in Survival Curves," Transmission of Viruses
     by the Water Route, Interscience Publishers, a division
     of John Wiley and Sons, New York, New York, pp. 235-240
     (1967) .

 5.  Clark, R. M., and Niehaus, J. F., "A Mathematical Model
     for Viral Devitalization," Transmission of Viruses by
     the Water Route, Interscience Publishers,a division of
     John Wiley and Sons, New York, New York, pp. 241-245 (1967).

 6.  Clark, R. M., "A Mathematical Model of the Kinetics of
     Viral Devitalization," Mathematical Biosciences 2, pp.  413-423
     (1968) .

 7.  Willers, A., FR, Practical Analysis, Dover, New York (1948).

 8.  Hildebrand, F. B., Introduction to Numerical Analysis,
     McGraw-Hill, New York, New York (1956).

 9.  Draper, N., and Smith, H., Applied Regression Analysis,
     John Wiley and Sons, New York, New York (1967).

10.  Sharp, G. D., "Electron Microscopy and Viral Particle
     Function," Transmission of Viruses by the Water Route,
     Interscience Publishers, a division of John Wiley and
     Sons, New York, New York, pp. 193-217  (19671.
                              34

-------
                              TECHNICAL REPORT DATA
                        (Pease read Instructions on the reverse before completing)
 . REPORT NO.
  EPA-670/2-74-067
2.
                         3. RECIPIENT'S ACCESSION'NO.
4. TITLE AND SUBTITLE

  A MATHEMATICAL ANALYSIS  OF THE KINETICS
  OF VIRAL  INACTIVATION
                         5. REPORT DATE
                         August 1974; Issuing Date
                         6. PERFORMING ORGANIZATION CODE
 '. AUTHOR(S)          ~~     ~~~~'	
  Robert M.  Clark, Betty  Lou Grupenhoff,
  and George C.  Kent
                         8. PERFORMING ORGANIZATION REPORT NO.
 9. PERFORMING ORG \NIZATION NAME AND ADDRESS
  National  Environmental  Research Center
  Office  of Research and  Development
  U.S. Environmental Protection Agency
  Cincinnati,  Ohio  45268
                         1O. PROGRAM ELEMENT NO. T.CB047 '
                         ROAP  21AQE; Task 10
                         11. CONTRACT/GRANT NO.
 12. SPONSORING AGENCY NAME AND ADDRESS
  Same as  above
                                                   13. TYPE OF REPORT AND PERIOD COVERED
                                                   14. SPONSORING AGENCY CODE
 15. SUPPLEMENTARY NOTES
 16. ABSTRACT          •

  Pathogenic enteric viruses transmitted via the water route present
  a potential hazard to  public health because of their resistance to
  natural  or artificial  disinfection mechanisms.  Of constant concern
  to public health officials is the ability of viruses to  pass through
  water  treatment plants.   Therefore, many research investigations have
  been directed toward the study of the  inactivation of viruses and
  enteric  organisms.  This report describes a mathematical model which
  can be used to characterize the response of viruses to a disinfecting
  agent.   Not only is the  model presented, but a technique is described
  which  can be used to estimate the model's parameters.  Both the model
  and the  estimation technique are being used to analyze experimental
  information resulting  from disinfection studies.
 17.
                            KEY WORDS AND DOCUMENT ANALYSIS
                DESCRIPTORS
                                        b.lDENTIFIERS/OPEN ENDED TERMS
                                     c. COS AT I Field/Group
  Computation,  Computers,  *Mathe-
  matical models, *Viruses,  Linear
  regression, Disinfection,  *Electron
  microscopy
               Exponential  decay,
               *Inactivation
    12A
    13B
 8. DISTRIBUTION STATEMENT
              19. SECURITY CLASS (ThisReport)'
                  UNCLASSIFIED
21. NO. OF PAGES
    39
         RELEASE TO PUBLIC
              20. SECURITY CLASS (This page)
                  UNCLASSIFIED
                                                               22. PRICE
EPA Form 2220-1 (9-73)
                                                           VS. GOVERNMENT PRINTING OFFICE: 1974- 657-049/1025

-------