Systems and methods for step discontinuity removal in real-time PCR fluorescence data

Patent metadata

Field	Value
Publication number	US-10133843-B2
Application number	US-201213688133-A
Country	US
Kind code	B2
Filing date	Nov 28, 2012
Priority date	May 13, 2008
Publication date	Nov 20, 2018
Grant date	Nov 20, 2018

Abstract

Systems and methods for removing jump discontinuities in growth data are provided. A first approximation to a received data set is determined by applying a non-linear regression process to a non-linear function that models the data set to determine parameters, including a step discontinuity parameter. A second approximation to the data set is also determined by applying a regression process to a second non-linear function to determine parameters, including a step discontinuity parameter, of the second function. One of the approximations is selected based on an information coefficient determined for each of the approximations. If a confidence interval for the step discontinuity parameter includes zero, no correction is made, and if includes zero, then a correction is made. For a correction, the portion of the data curve prior to the step change is replaced with appropriate portion of the selected approximation to produce a shift-corrected data set.

First claim

What is claimed is:

1. A method of automatically removing a step discontinuity in data representing a Polymerase Chain Reaction (PCR) growth process, the method comprising:
detecting, with a PCR system, intensity values of signals generated using a biological and/or chemical reaction sample undergoing the PCR growth process, at a plurality of growth cycles of the PCR growth process;
receiving, by a computer system communicatively coupled to the PCR system, a dataset representing the PCR growth process, the dataset including a plurality of data points, each data point having a pair of coordinate values, wherein the pair of coordinate values correspond to a cycle number of the PCR growth process and an intensity value of a signal generated using the biological and/or chemical reaction sample and detected by the PCR system after a growth cycle corresponding to the cycle number;
calculating, by the computer system, a first approximation of a curve that fits the dataset by applying a first non-linear regression process to a first non-linear function to determine parameters of the first non-linear function, said parameters including a first step discontinuity parameter that represents a magnitude of the step discontinuity, the first non-linear function including the first step discontinuity parameter multiplying a step function;
calculating, by the computer system, a second approximation of a curve that fits the dataset by applying a second regression process to a second non-linear function to determine parameters of the second non-linear function, said parameters of the second non-linear function including a second step discontinuity parameter that represents the magnitude of the step discontinuity, the second non-linear function including the second step discontinuity parameter multiplying a step function, wherein the second non-linear function is different from the first non-linear function;
determining, by the computer system, an information coefficient for each of the first and second approximations, each information coefficient indicating an accuracy of a fit of the corresponding approximation to the dataset;
selecting, by the computer system, one of the approximations based on the information coefficient, the selected approximation providing a best fit to the dataset;
determining, by the computer system, a confidence interval of the step discontinuity parameter for the selected approximation;
if the confidence interval does not include the value zero, replacing, by the computer system, a portion of the dataset with the selected approximation with the corresponding step discontinuity parameter set to zero, wherein the method is implemented in a computer system having a processor; and
determining, by the computer system, a quantity of a target molecule in the biological and/or chemical reaction sample by determining a cycle threshold (C t ) value using the dataset.

2. The method of claim 1 , wherein the first non-linear regression process is a Levenberg-Marquardt (LM) regression process and wherein the first non-linear function is a double sigmoid function.

3. The method of claim 2 , wherein the double sigmoid function is of the form:

a
+

b
·
x

+

c

(

1
+

exp
⁡

(

-
d

·

(

x
-
e

)

)

)

⁢

(

1
+

exp
⁡

(

-
f

·

(

x
-
g

)

)

)

+

h
·

(

UnitStep
⁡

(
x
)

-

UnitStep
⁡

(

x
-
cac

)

)

,

wherein UnitStep is of the form:

UnitStep
⁡

(
x
)

=

{

1

x
≥
0

0

x
<
0

,

wherein h is the first step discontinuity parameter, and
wherein calculating the first approximation includes iteratively determining one or more of the parameters a, b, c, d, e, f, g and h of the double sigmoid function.

4. The method of claim 1 , wherein the portion of the dataset replaced includes the portion of t

Assignees

• Roche Molecular Systems Inc

Inventors

• Sane Aditya P
• Kurnik Ronald T
• Baldanza Jonathan M

Classifications

• G06F2218/10

  by analysing the shape of a waveform, e.g. extracting parameters relating to peaks · CPC title

• G06F2218/02

  Preprocessing · CPC title

• C12Q2537/165

  Mathematical modelling, e.g. logarithm, ratio · CPC title

• C12Q1/6851

  Quantitative amplification · CPC title

• G06F19/22 Primary

  Physics · mapped topic