pcrfit_single: pcrfit_single - A function to extract features from an...

In PCRedux: Quantitative Polymerase Chain Reaction (qPCR) Data Mining and Machine Learning Toolkit

pcrfit_single - A function to extract features from an amplification curve

Description

The pcrfit_single is responsible for the extraction of features from amplification curve data. The function can be used for custom functions for a paralleled analysis of amplification curve data. An example is given in the vignette.

Usage

pcrfit_single(x)

Arguments

x	is the data set containing the fluorescence amplitudes.

Details

Details can be found in the vignette.

Value

Output Description

"cpD1"	maximum of the first derivative curve	numeric
"cpD2"	maximum of the second derivative curve	numeric
"cpD2_approx"	maximum of the second derivative curve calculated by the approximate derivative	numeric
"cpD2_ratio"	a value calculated from the ratio between cpD2 and cpD2_approx	numeric
"eff"	qPCR amplification efficiency	numeric
"sliwin"	qPCR amplification efficiency according the the 'window-of-linearity' method by Ruijter et al. (2009)	numeric
"cpDdiff"	absolute difference between cpD1 and cpD2	numeric
"loglin_slope"	slope determined by a linear model of the data points from the minimum and maximum of the second derivative	numeric
"cpD2_range"	cycle difference between the maximum and the minimum of the second derivative curve	numeric
"top"	takeoff point. When no top can be determined, the tob value is set to the first cycle number.	numeric
"f.top"	fluorescence at takeoff point. When no f.tdp can be determined, the f.tdp value is set to the RFU value at the first cycle number.	numeric
"tdp"	takes the maximum fluorescence subtracted by reverse values of the fluorescence and calculates then the fake takeoff point. It is so to speak the take down point (tdp). When no tdp can be determined, the tdb value is set to the last cycle number.	numeric
"f.tdp"	fluorescence at tdp point. When no f.tdp can be determined, the f.tdp value is set to the RFU value at the last cycle number.	numeric
"bg.stop"	estimates the end (cycle) the amplification curve background based on the bg.max function and normalizes it to the total cycle number	numeric
"amp.stop"	estimates the end (cycle) of the amplification curve based in the bg.max function and normalizes it to the total cycle number	numeric
"b_slope"	Is the slope of the seven parameter model	numeric
"b_model_param"	Is the b model parameter of the model optimally fitted according to the AIC	numeric
"c_model_param"	Is the c model parameter of the model optimally fitted according to the AIC	numeric
"d_model_param"	Is the d model parameter of the model optimally fitted according to the AIC	numeric
"e_model_param"	Is the e model parameter of the model optimally fitted according to the AIC	numeric
"f_model_param"	Is the f model parameter of the model optimally fitted according to the AIC	numeric
"f_intercept"	Is the intercept of the seven parameter model	numeric
"convInfo_iteratons"	Number of iterations needed to fit the 7 parameter model	numeric
"qPCRmodel"	non-linear model determined for the analysis	factor
"qPCRmodelRF"	non-linear model determined for the analysis of the reversed amplification curve	factor
"minRFU"	minimum of fluorescence amplitude	numeric
"maxRFU"	maximum of fluorescence amplitude	numeric
"init2"	initial template fluorescence from an exponential model	numeric
"fluo"	raw fluorescence value at the point defined by cpD2	numeric
"slope_bg"	slope of the first cycles	numeric
"k1_model_param"	Is the k1 model parameter of the seven parameter model	numeric
"k2_model_param"	Is the k2 model parameter of the seven parameter model	numeric
"intercept_bg"	intercept of the first cycles	numeric
"sigma_bg"	sigma of background	numeric
"sd_bg"	standard deviation of the background (ground phase) region (start to takeoff point)	numeric
"head2tail_ratio"	ratio between the signal of the background and tail region	numeric
"mblrr_intercept_bg"	the value of the intercept in the estimated background region of the amplification curve	numeric
"mblrr_slope_bg"	the value of the slope in the estimated background region of the amplification curve	numeric
"mblrr_cor_bg"	the value of the linear correlation coefficient in the estimated background region of the amplification curve	numeric
"mblrr_intercept_pt"	the value of the intercept in the estimated plateau phase of the amplification curve	numeric
"mblrr_slope_pt"	the value of the slope in the estimated plateau phase of the amplification curve	numeric
"mblrr_cor_pt"	the value of the linear correlation coefficient in the estimated plateau phase of the amplification curve	numeric
"polyarea"	area of a polygon given by the vertices in the vectors cycles and fluorescence	numeric
"peaks_ratio"	Takes the estimate approximate local minimums and maximums
"autocorrelation"	is a value of autocorrelation of a gain curve from a quantitative PCR experiment	numeric
"cp_e.agglo"	agglomerative hierarchical estimate for multiple change points	numeric
"cp_bcp"	change point by Bayesian analysis methods	numeric
"amptester_shapiro"	tests based on the Shapiro-Wilk normality test if the amplification curve is just noise	binary
"amptester_lrt"	performs a cycle dependent linear regression and determines if the coefficients of determination deviates from a threshold	binary
"amptester_rgt"	Resids growth test (RGt) tests if fluorescence values in a linear phase are stable	binary
"amptester_tht"	Threshold test (THt) takes the first 20 percent and the last 15 percent of any input data set and performs a Wilcoxon rank sum tests.	binary
"amptester_slt"	Signal level test compares 1. the signals by a robust "sigma" rule by median + 2 * mad and 2. by comparison of the signal/noise ratio	binary
"amptester_polygon"	pco test (pco) determines if the points in an amplification curve (like a polygon, in particular non-convex polygons) are in a "clockwise" order.	binary
"amptester_slope.ratio"	SlR uses the inder function to find the approximated first derivative maximum, second derivative minimum and the second derivative maximum. These are used for a regression analysis with the corresponding fluorescence amplitude data.	numeric
"hookreg_hook"	estimate of hook effect like curvature	binary
"hookreg_hook_slope"	estimate of slope of the hook effect like curvature	numeric
"hookreg_hook_delta"	Estimated value for the number of cycles from the qPCR cycle where the hook effect was determined up to the last qPCR cycle	numeric
"central_angle"	shows the central angle calculated from the maximum and minimum of the second derivatives, with the first derivative maximum being the center	numeric
"sd_bg"	shows the standard deviation of the fluorescence in the ground phase	numeric
"number_of_cycles"	Number of cylces	numeric
"direction"	test if the maximum of the first derivative is positive or negative	numeric
"range"	outputs the difference of fluorescence between 0.99 and 0.01 percentile. The value thus corresponds approximately to the maximum achievable signal difference of an amplification curve.	numeric
"polyarea_trapz"	calculates trapezoidal integration. The calculation stops when the difference from one step to the next is smaller than a tolerance value, or the iterations become too large. The value corresponds to the sum signal via the total amplification curve.	numeric
"cor"	is the value of the correlation coefficient from a linear correlation analysis according to Pearson between all PCR cycles and the fluorescence signals.	numeric
"res_coef_pcrfit.b"	is the parameter from the adjustment with a nonlinear (sigmoid) four-parametric model which describes the Hill’s slope of the curve (i.e. this is related to the steepness of the curve point e)	numeric
"res_coef_pcrfit.c"	is the parameter from the adjustment with a nonlinear (sigmoid) four-parametric model which describes the maximum value that can be obtained (i.e. what happens at infinite number of cycles)	numeric
"res_coef_pcrfit.d"	is the parameter from the adjustment with a nonlinear (sigmoid) four-parametric model which describes the minimum value that can be obtained (i.e. what happens at 0 cycles)	numeric
"res_coef_pcrfit.e"	is the parameter from the adjustment with a nonlinear (sigmoid) four-parametric model, which describes the point of inflection (i.e. the point on the sigmoid curve halfway between d and c)	numeric
"fitAIC"	is the value of the Akaike's second-order corrects Information Criterion, which was determined on a non-linear (sigmoid) four-parameter model	numeric
"fitIter"	Number of iterations needed to fit the 4 parameter model	numeric
"segment_x"	Adjusts a regression model with segmented (linear) relationships between fluorescence and PCR cycles. This segment describes the baseline in an amplification curve.	numeric
"segment_U1.x"	Adjusts a regression model with segmented (linear) relationships between fluorescence and PCR cycles. This segment describes the slope in an amplification curve.	numeric
"segment_U2.x"	Adjusts a regression model with segmented (linear) relationships between fluorescence and PCR cycles. This segment describes the plateau in an amplification curve.	numeric
"segment_psi1.x"	Adjusts a regression model with segmented (linear) relationships between fluorescence and PCR cycles. The value is based on the break-point(s) fixed at the values and describes the transition from the baseline phase to the exponential phase.	numeric
"segment_psi2.x"	Adjusts a regression model with segmented (linear) relationships between fluorescence and PCR cycles. The value is based on the break-point(s) fixed at the values and describes the transition from the exponential phase to the plateau phase.	numeric
"sumdiff"	describes proportion of cycles x in which the fluorescence signal of x is smaller than in x+1	numeric
"poly_1"	is a value of a third-order polynomial a + b*x + c*x^2 + d*x^3 is fitted to the curve data, where the intercept correspond to the baseline and the three predictor-dependent terms deliver a approximation to the sigmoidal curve structure. This value describes the intercept "a" (cutting point of the amplification curve model with the ordinate).	numeric
"poly_2"	is a value of a third-order polynomial a + b*x + c*x^2 + d*x^3 is fitted to the curve data, where the intercept correspond to the baseline and the three predictor-dependent terms deliver a approximation to the sigmoidal curve structure. This value describes the linear part "b*x".	numeric
"poly_3"	is a value of a third-order polynomial a + b*x + c*x^2 + d*x^3 is fitted to the curve data, where the intercept correspond to the baseline and the three predictor-dependent terms deliver a approximation to the sigmoidal curve structure. This value describes the quadratic part "c*x^2".	numeric
"poly_4"	is a value of a third-order polynomial a + b*x + c*x^2 + d*x^3 is fitted to the curve data, where the intercept correspond to the baseline and the three predictor-dependent terms deliver a approximation to the sigmoidal curve structure. This value describes the cubic part "d*x^3".	numeric
"window_Win_1"	The complete curve trajectory is segmented into 10 equidistant windows by fitting an interpolating smoothing spline with smoothing factor 0.5 to the curve, interpolating exactly 50 curve points, and then cutting these into 10 windows of five values each, with a subsequent calculation of the MAD/Median ratio for each of these windows. This is the 1st window for cycles (0.961,4.9].	numeric
"window_Win_2"	The complete curve trajectory is segmented into 10 equidistant windows by fitting an interpolating smoothing spline with smoothing factor 0.5 to the curve, interpolating exactly 50 curve points, and then cutting these into 10 windows of five values each, with a subsequent calculation of the MAD/Median ratio for each of these windows. This is the 2nd window for cycles (4.9,8.8].	numeric
"window_Win_3"	The complete curve trajectory is segmented into 10 equidistant windows by fitting an interpolating smoothing spline with smoothing factor 0.5 to the curve, interpolating exactly 50 curve points, and then cutting these into 10 windows of five values each, with a subsequent calculation of the MAD/Median ratio for each of these windows. This is the 3rd window for cycles (8.8,12.7].	numeric
"window_Win_4"	The complete curve trajectory is segmented into 10 equidistant windows by fitting an interpolating smoothing spline with smoothing factor 0.5 to the curve, interpolating exactly 50 curve points, and then cutting these into 10 windows of five values each, with a subsequent calculation of the MAD/Median ratio for each of these windows. This is the 4th window for cycles (12.7,16.6].	numeric
"window_Win_5"	The complete curve trajectory is segmented into 10 equidistant windows by fitting an interpolating smoothing spline with smoothing factor 0.5 to the curve, interpolating exactly 50 curve points, and then cutting these into 10 windows of five values each, with a subsequent calculation of the MAD/Median ratio for each of these windows. This is the 5th window for cycles (16.6,20.5].	numeric
"window_Win_6"	The complete curve trajectory is segmented into 10 equidistant windows by fitting an interpolating smoothing spline with smoothing factor 0.5 to the curve, interpolating exactly 50 curve points, and then cutting these into 10 windows of five values each, with a subsequent calculation of the MAD/Median ratio for each of these windows. This is the 6th window for cycles (20.5,24.4].	numeric
"window_Win_7"	The complete curve trajectory is segmented into 10 equidistant windows by fitting an interpolating smoothing spline with smoothing factor 0.5 to the curve, interpolating exactly 50 curve points, and then cutting these into 10 windows of five values each, with a subsequent calculation of the MAD/Median ratio for each of these windows. This is the 7th window for cycles (24.4,28.3].	numeric
"window_Win_8"	The complete curve trajectory is segmented into 10 equidistant windows by fitting an interpolating smoothing spline with smoothing factor 0.5 to the curve, interpolating exactly 50 curve points, and then cutting these into 10 windows of five values each, with a subsequent calculation of the MAD/Median ratio for each of these windows. This is the 8th window for cycles (28.3,32.2].	numeric
"window_Win_9"	The complete curve trajectory is segmented into 10 equidistant windows by fitting an interpolating smoothing spline with smoothing factor 0.5 to the curve, interpolating exactly 50 curve points, and then cutting these into 10 windows of five values each, with a subsequent calculation of the MAD/Median ratio for each of these windows. This is the 9th window for cycles (32.2,36.1].	numeric
"window_Win_10"	The complete curve trajectory is segmented into 10 equidistant windows by fitting an interpolating smoothing spline with smoothing factor 0.5 to the curve, interpolating exactly 50 curve points, and then cutting these into 10 windows of five values each, with a subsequent calculation of the MAD/Median ratio for each of these windows. This is the 10th window for cycles (36.1,40].	numeric
"sd_plateau"	describes the standard deviation in the late phase of an amplification curve (last five cycles). With ideal PCRs, this corresponds to the plateau phase.	numeric

gives a data.frame (S3 class, type of list) as output for the curve features

Author(s)

Stefan Roediger, Michal Burdukiewcz

References

M. Febrero-Bande, M.O. de la Fuente, others, Statistical computing in functional data analysis: The R package fda.usc, Journal of Statistical Software. 51 (2012) 1–28. http://www.jstatsoft.org/v51/i04/

A.-N. Spiess, C. Deutschmann, M. Burdukiewicz, R. Himmelreich, K. Klat, P. Schierack, S. Roediger, Impact of Smoothing on Parameter Estimation in Quantitative DNA Amplification Experiments, Clinical Chemistry. 61 (2015) 379–388. doi:10.1373/clinchem.2014.230656.

S. Roediger, A. Boehm, I. Schimke, Surface Melting Curve Analysis with R, The R Journal. 5 (2013) 37–53. http://journal.r-project.org/archive/2013-2/roediger-bohm-schimke.pdf.

S. Roediger, M. Burdukiewicz, K.A. Blagodatskikh, P. Schierack, R as an Environment for the Reproducible Analysis of DNA Amplification Experiments, The R Journal. 7 (2015) 127–150. http://journal.r-project.org/archive/2015-1/RJ-2015-1.pdf.

S. Pabinger, S. Roediger, A. Kriegner, K. Vierlinger, A. Weinhauusel, A survey of tools for the analysis of quantitative PCR (qPCR) data, Biomolecular Detection and Quantification. 1 (2014) 23–33. doi:10.1016/j.bdq.2014.08.002.

S. Roediger, M. Burdukiewicz, P. Schierack, chipPCR: an R package to pre-process raw data of amplification curves, Bioinformatics. 31 (2015) 2900–2902. doi:10.1093/bioinformatics/btv205.

See Also

bcp bg.max,amptester,smoother e.agglo diffQ,mcaPeaks,diffQ2 head2tailratio,earlyreg,hookreg,hookregNL,mblrr,autocorrelation_test polyarea pcrfit,takeoff,sliwin,efficiency diff quantile segmented

Examples

# Load the chipPCR package and analyze from the C126EG685 the first qPCR run
# "A01" (column 2).
library(chipPCR)
res <- pcrfit_single(C126EG685[, 2])

PCRedux documentation built on May 11, 2022, 5:18 p.m.