Multicomponent analysis method including the determination of a statistical confidence interval

Claim 1

A signal processing method in a processor for performing a multicomponent analysis of a signal resulting from a spectral response of a mixture comprising a plurality of spectrally resolvable molecular species, including the determination of a concentration estimate and a statistical confidence interval for each species, such method comprising the steps of: receiving a data vector d from a multichannel detector, data vector d having length n c , n c being the number of detector channels being monitored, the magnitude of each vector element d i representing a signal intensity at a particular channel i; providing a calibration matrix K having n c rows and n p columns wherein n c is larger than n p , n p being the number of spectrally resolvable molecular species, the magnitude of each matrix element K ij representing a calibration value at a particular channel i for a particular species j; determining a concentration estimate vector c having length n p , the magnitude of each vector element c i representing a concentration estimate of a particular species i; determining a confidence interval Ci i for each of the elements of the concentration estimate vector according to the expression CI.sub.i =c.sub.i ±(varcovar(c.sub.ii)).sup.1/2 Q.sub.(cl,nf) where Q is a critical value of a statistical distribution for a given level of confidence, cl, and a given number of degrees of freedom n f , where n f =n c -n p .

Claim 14

A method of polynucleotide sequencing comprising the steps of: (a) forming a mixture of a first, a second, a third, and a forth class of labeled polynucleotides such that each polynucleotide in the first class includes a 3'-terminal dideoxyadenosine and is labeled with a first label, each polynucleotide in the second class includes a 3'-terminal dideoxycytidine and is labeled with a second label, each polynucleotide in the third class includes a 3'-terminal dideoxyguanosine and is labeled with a third label, and each polynucleotide in the forth class includes a 3'-terminal dideoxythymidine and is labeled with a forth label, wherein each of the first, second, third, and fourth labels are spectrally resolved; (b) electrophoretically separating the labeled polynucleotides thereby forming multiple bands of similarly sized polynucleotides; (c) detecting the bands using a spectral array detector having n c channels, where n c is greater than four, (d) providing a calibration matrix K having n c rows and n p columns wherein n c is larger than n p , n p being the number of spectrally resolvable molecular species, the magnitude of each matrix element K ij representing a calibration value at a particular channel i for a particular species j; (e) determining a concentration estimate vector c having length n p , the magnitude of each vector element c i representing a concentration estimate of a particular species i; (f) determining a confidence interval Ci i for each of the elements of the concentration estimate vector according to the expression CI.sub.i =c.sub.i ±(varcovar(c.sub.ii)).sup.1/2 Q.sub.(cl,nf) where Q is a critical value of a statistical distribution for a given level of confidence, cl, and a given number of degrees of freedom n f , where n f =n c -n p ; (g) reporting values of c i and CI i ; and (i) repeating steps (c)-(g) until all bands of interest have been analyzed.

Claim 15

A method of polynucleotide sequencing comprising the steps of: (a) forming a mixture of a first, a second, a third, and a forth class of labeled polynucleotides such that each polynucleotide in the first class includes a 3'-terminal dideoxyadenosine and is labeled with a first label, each polynucleotide in the second class includes a 3'-terminal dideoxycytidine and is labeled with a second label, each polynucleotide in the third class includes a 3'-terminal dideoxyguanosine and is labeled with a third label, and each polynucleotide in the forth class includes a 3'-terminal dideoxythymidine and is labeled with a forth label, wherein each of the first, second, third, and fourth labels are spectrally resolved; (b) electrophoretically separating the labeled polynucleotides thereby forming multiple bands of similarly sized polynucleotides; (c) detecting the bands using a spectral array detector having n c channels, where n c is greater than four, (d) providing a calibration matrix K having n c rows and n p columns wherein n c is larger than n p , n p being the number of spectrally resolvable molecular species, the magnitude of each matrix element K ij representing a calibration value at a particular channel i for a particular species j; (e) determining a concentration estimate vector c having length n p , the magnitude of each vector element c i representing a concentration estimate of a particular species i; (f) determining a confidence interval Ci i for each of the elements of the concentration estimate vector according to the expression CI.sub.i =c.sub.i ±(varcovar(c.sub.ii)).sup.1/2 Q.sub.(cl,nf) where Q is a critical value of a statistical distribution for a given level of confidence, cl, and a given number of degrees of freedom n f , where n f =n c -n p ; (g) reporting values of c i and CI i ; and (i) repeating steps (c)-(g) until all bands of interest have been analyzed.

Claim 16

An apparatus for measuring the spectral response of a sample including a mixture containing n p spectrally distinguishable species comprising: a spectral array detector for detecting the spectral response of the sample, the detector having n c channels where n c is greater than n p , the output of the detector being a signal representing a light intensity at a particular channel at a particular time; a signal processor for receiving a signal from the spectral array detector, the signal processor being configured to perform the following steps: receiving a data vector d from the spectral array detector, data vector d having length n c , n c being the number of detector channels being monitored, the magnitude of each matrix element d i representing a signal intensity at a particular channel i; providing a calibration matrix K having n c rows and n p columns wherein n c is larger than n p , n p being the number of spectrally resolvable molecular species, the magnitude of each matrix element K ij representing a calibration value at a particular channel i for a particular species j; determining a concentration estimate vector c having length n p , the magnitude of each matrix element c i representing a concentration estimate of a particular species i; determining a confidence interval for each of the elements of the concentration estimate vector CI i according to the expression CI.sub.i =c.sub.i ±(varcovar(c.sub.ii)).sup.1/2 Q.sub.(cl,nf) where Q is the critical value of a statistical distribution for a given level of confidence, cl, and a given number of degrees of freedom n f where n f =n c -n p .

Claim 17

An apparatus for measuring the spectral response of a sample including a mixture containing n p spectrally distinguishable species comprising: a spectral array detector for detecting the spectral response of the sample, the detector having n c channels where n c is greater than n p , the output of the detector being a signal representing a light intensity at a particular channel at a particular time; a signal processor for receiving a signal from the spectral array detector, the signal processor being configured to perform the following steps: receiving a data vector d from the spectral array detector, data vector d having length n c , n c being the number of detector channels being monitored, the magnitude of each matrix element d i representing a signal intensity at a particular channel i; providing a calibration matrix K having n c rows and n p columns wherein n c is larger than n p , n p being the number of spectrally resolvable molecular species, the magnitude of each matrix element K ij representing a calibration value at a particular channel i for a particular species j; determining a concentration estimate vector c having length n p , the magnitude of each matrix element c i representing a concentration estimate of a particular species i; determining a confidence interval for each of the elements of the concentration estimate vector CI i according to the expression CI.sub.i =c.sub.i ±(varcovar(c.sub.ii)).sup.1/2 Q.sub.(cl,nf) where Q is the critical value of a statistical distribution for a given level of confidence, cl, and a given number of degrees of freedom n f where n f =n c -n p .