CN101408501A - Method for quantitatively detecting DNA base by using near-infrared spectrum-partial least squares method - Google Patents

Abstract

The invention relates to a method for quantitative detection of DNA bases by adopting a near infrared spectrum-partial least squares method, belonging to the technical field of near infrared spectrum detection. The present high performance liquid chromatography and the test equipment thereof are expensive, the cost is higher, the operation is more complex, the testing time is longer, and the variability of the test data is larger. The invention adopts the near infrared spectrum-partial least squares method to detect the contents of four kinds of bases of DNA, and the method comprises the following steps: step 1, the near infrared diffuse reflection spectrum of the DNA base mixture is detected; and step 2, the detected near infrared diffuse reflection spectrum is analyzed by a partial least squares method. A DNA base quantitative analysis model is established by adopting the partial least squares method, and firstly, the spectral pretreatment is carried out on the detected near infrared diffuse reflection spectrum by adopting a first derivative spectrum method; then the optimum main component is selected by adopting an interactive detection method, and the predicted square sum of residuals is taken as the evaluation criterion for verification; and then all the samples for modeling are predicted by using the quantitative analysis model, and the difference between the predicted value and the known reference value of each sample is calculated. The invention is suitable for the detection of the DNA base.

Description

Adopt the method for near infrared spectrum-partial least square method detection by quantitative DNA base

Technical field

(Partial Least Squares, PLS) the combine method of detection by quantitative DNA base belongs to the Near Infrared Spectroscopy Detection Technology field to the present invention relates to a kind of near infrared spectrum (NIR) method and partial least square method.

Background technology

Adenine (A), guanine (G), cytimidine (C) and thymine (T) are the important component parts of DNA, and more to its separation and analytical approach, that generally adopt at present is high performance liquid chromatography (HPLC).This method is according to different separation and the analyses that realize the DNA base with the partition factor between the stationary phase of the moving phase (leacheate) of sample fraction in chromatographic column.After sample entered in the chromatographic column along with moving phase, component was with regard to therein two alternate repeated multiple times (10 of carrying out ³～10 ⁶) distribute i.e. adsorption-desorption-emit.Because stationary phase is to the adsorptive power difference of various components, i.e. preservation effect difference, therefore, the travelling speed of each component in chromatographic column is just different, and be through behind certain column length, just separated from one another, order is left chromatographic column and is entered detecting device, the ion flow signal that produces is depicted the chromatographic peak of each component, thereby is obtained each components contents after amplifying on register.High performance liquid chromatography has that speed is fast, efficient is high, the characteristics of highly sensitive, operation automation.

Near-infrared spectrum wavelength is distributed in 1100～2500nm, and near infrared spectroscopy is the method for complicated chemical composition in the working sample.Its spectral characteristic is stable, have sample pre-treatments simple and direct, need not chemical reagent, environmental protection, simple to operate, detection speed fast, good stability and can be implemented in advantage such as line analysis.In the method, the generation of near infrared spectrum mainly is because the anharmonicity of molecular vibration.In near infrared spectral range, frequency multiplication and the sum of fundamental frequencies of measuring hydrogeneous radicals X-H (X=C, N, O, S etc.) vibration absorb.Because frequency multiplication and sum of fundamental frequencies transition probability are low, and organic substance is the absorption of frequency multiplication and sum of fundamental frequencies in the near infrared spectrum district, so a little less than the extinction coefficient, bands of a spectrum are overlapping serious.Therefore, the useful information that extracts near infrared spectrum belongs to weak information and multiple information, finishes qualitative and quantitative analysis by means of chemometrics method such as partial least square method usually.

Summary of the invention

Its checkout equipment costliness of high performance liquid chromatography of the prior art, expense is higher, and operation is complicated, test consuming time longlyer, and test figure makes a variation bigger.And have not yet to see combine with the partial least square method report of detection by quantitative DNA base of near infrared spectroscopy.The method that the objective of the invention is to combine with partial least square method with near infrared spectroscopy replaces high performance liquid chromatography, easy detection by quantitative DNA base fast,, for this reason, we have invented a kind of method that adopts near infrared spectrum-partial least square method detection by quantitative DNA base.

The present invention's method is characterised in that, adopts near infrared spectrum-partial least square method to detect the content of four kinds of bases of DNA, and its step is as follows:

1, detects the near-infrared diffuse reflection spectrum of DNA base mixture

Sample is carried out near infrared scanning, and the scanning wavelength scope is 1100～2500nm;

2, analyze detected near-infrared diffuse reflection spectrum with partial least square method

(1) adopts partial least square method to set up DNA base Quantitative Analysis Model, that is: at first adopt first-derivative spectroscopy measured infrared diffuse light stalking row spectrum pre-service; Then, adopt the cross-verification method to select best number of principal components, verify as evaluation criterion with prediction residual quadratic sum (PRESS);

(2) with described Quantitative Analysis Model each sample of participating in modeling is predicted, obtained the poor of the predicted value of each sample and known reference value.

Its technique effect of the present invention is, sets up DNA base Quantitative Analysis Model, thereby does not need sample is carried out chemical treatment, and its precision of prediction can satisfy the requirement of the quantitative test that detects micro-example, and detection speed is fast, simple to operate.Measured infrared diffuse reflectance spectroscopy is overlapping serious, adopts common spectroscopic analysis methods to be difficult to carry out quantitative test, and adopts the offset minimum binary rule can realize analyzing.The selection of best number of principal components is directly connected to the actual prediction ability of DNA base Quantitative Analysis Model, number of principal components is very few, just can not fully reflect sample spectra information, number of principal components is crossed at most can also mix calculating with the information of some noises, reduce the predictive ability of model, and adopt the cross-verification method, as evaluation criterion, determined required best number of principal components with prediction residual quadratic sum (PRESS).

Description of drawings

Fig. 1 is the pre-service original spectrum of four kinds of DNA base mixtures.Fig. 2 is the pre-processed spectrum first order derivative curve map of four kinds of DNA base mixtures, and this figure double as is a Figure of abstract.Fig. 3 is the graph of a relation of adenine number of principal components and PLS Quantitative Analysis Model PRESS value.Fig. 4 is the graph of a relation of thymine number of principal components and PLS Quantitative Analysis Model PRESS value.Fig. 5 is the graph of a relation of guanine number of principal components and PLS Quantitative Analysis Model PRESS value.Fig. 6 is the graph of a relation of cytimidine number of principal components and PLS Quantitative Analysis Model PRESS value.Fig. 7 is adenine normal concentration and prediction concentrations corresponding relation calibration set curve map.Fig. 8 is adenine normal concentration and forecast concentration corresponding relation forecast set curve map.Fig. 9 is thymine normal concentration and prediction concentrations corresponding relation calibration set curve map.Figure 10 is thymine normal concentration and prediction concentrations corresponding relation forecast set curve map.Figure 11 is guanine normal concentration and prediction concentrations corresponding relation calibration set curve map.Figure 12 is guanine normal concentration and prediction concentrations corresponding relation forecast set curve map.Figure 13 is cytimidine normal concentration and prediction concentrations corresponding relation calibration set curve map.Figure 14 is cytimidine normal concentration and prediction concentrations corresponding relation forecast set curve map.

Embodiment

The preparation of four kinds of DNA base mixture samples, to be divided into two groups at random by 35 four kinds DNA base mixture samples of variable concentrations proportioning, one group is calibration set (the calibration set), comprise 30 four kinds DNA base mixture samples, another group comprises 5 four kinds DNA base mixture samples for forecast set (the prediction set).Draw the normal concentration value of 35 samples according to recipe calculation.

Adopt near infrared spectrum-partial least square method to detect the content of four kinds of DNA bases, its step is as follows:

1, detects the near-infrared diffuse reflection spectrum of DNA base mixture

In the sample cell of integrating sphere, adopt Tianjin, island UV-3100 type UV, visible light near infrared spectrometer that sample is carried out near infrared scanning, spectral bandwidth 12nm base mixture sample compressing tablet, medium sweep, the scanning wavelength scope is 1100～2500nm, and each sample scanning 3 times is averaged.The near-infrared diffuse reflection spectrum of four kinds of measured DNA base mixtures is seen Fig. 1, shown in Figure 2.

2, analyze detected near-infrared diffuse reflection spectrum with partial least square method

(1) adopts partial least square method to set up DNA base Quantitative Analysis Model, that is: at first adopt first-derivative spectroscopy that measured infrared diffuse reflectance spectroscopy is carried out the spectrum pre-service; Then, adopt the cross-verification method to select best number of principal components, verify as evaluation criterion with prediction residual quadratic sum (PRESS);

(2) with described Quantitative Analysis Model each sample of participating in modeling is predicted, obtained the poor of the predicted value of each sample and known reference value.

The mathematic(al) representation of PRESS is:

$\mathrm{PRESS}=\underset{i=1}{\overset{n}{\mathrm{\Σ}}}\underset{j=1}{\overset{d}{\mathrm{\Σ}}}{(r_{p,\mathrm{ij}}-r_{\mathrm{ij}})}^2$

In the formula: n is a sample number in the calibration set; D sets up the number of principal components that model uses; r _{P, ij}It is the predicted value of sample; r _IjIt is the reference value of sample.

Determining of best number of principal components: when adopting partial least square method to set up DNA base Quantitative Analysis Model, the selection of number of principal components is directly connected to the actual prediction ability of DNA base Quantitative Analysis Model, number of principal components is very few, just abundant response sample spectral information; Number of principal components is crossed at most can also mix calculating with the information of some noises, reduces the predictive ability of model.When the PRESS value is more little, illustrate that the predictive ability of model is strong more, selected number of principal components is best.When the pairing four kinds of base number of principal components of selected spectrum were 4, model had minimum PRESS value; When number of principal components continued to increase, the PRESS value presented ascendant trend, sees shown in Fig. 3～6, thereby causes the model prediction ability drop, and therefore, the best number of principal components of sample is 4.

The method reliability is determined: after best number of principal components is determined, best DNA base Quantitative Analysis Model with the foundation of sample first derivative spectrum data, calibration set and forecast set sample concentration with four kinds of bases of this model prediction, the near infrared spectrum prediction concentrations of the calibration set of four kinds of bases and forecast set and the linear relationship of normal concentration are seen shown in Fig. 7～14, try to achieve during parameter that predicted value and standard value use and relative standard deviation (RSD) be listed in the table below, hence one can see that, and this method is reliable.

Claims (4)

1, a kind of method that adopts near infrared spectrum-partial least square method detection by quantitative DNA base is characterized in that, adopts near infrared spectrum-partial least square method to detect the content of four kinds of bases of DNA, and its step is as follows:

The first step, the near-infrared diffuse reflection spectrum of detection DNA base mixture carries out near infrared scanning to sample, and the scanning wavelength scope is 1100～2500nm;

In second step, analyze detected near-infrared diffuse reflection spectrum with partial least square method:

(1) adopts partial least square method to set up DNA base Quantitative Analysis Model, that is: at first adopt first-derivative spectroscopy that measured infrared diffuse reflectance spectroscopy is carried out the spectrum pre-service; Then, adopt the cross-verification method to select best number of principal components, verify as evaluation criterion with prediction residual quadratic sum (PRESS);

(2) with described Quantitative Analysis Model each sample of participating in modeling is predicted, obtained the poor of the predicted value of each sample and known reference value.

2, the method for detection by quantitative DNA base according to claim 1, it is characterized in that, the process for preparation of four kinds of DNA base mixture samples is, to be divided into two groups at random by 35 four kinds DNA base mixture samples of variable concentrations proportioning, one group is calibration set, comprise 30 four kinds DNA base mixture samples, another group comprises 5 four kinds DNA base mixture samples for forecast set; Draw the normal concentration value of 35 samples according to recipe calculation.

3, the method for detection by quantitative DNA base according to claim 1 is characterized in that, PRESS meets the following formula requirement:

$\mathrm{PRESS}=\underset{i=1}{\overset{n}{\mathrm{\Σ}}}\underset{j=1}{\overset{d}{\mathrm{\Σ}}}{(r_{p,\mathrm{ij}}-r_{\mathrm{ij}})}^2,$

In the formula: n is a sample number in the calibration set; D sets up the number of principal components that model uses; r _{P, ij}It is the predicted value of sample; r _IjIt is the reference value of sample.

4, the method for detection by quantitative DNA base according to claim 1 is characterized in that, best number of principal components is 4.

Images