Data processing: measuring – calibrating – or testing – Measurement system in a specific environment – Chemical analysis
Reexamination Certificate
2001-11-15
2004-01-06
Barlow, John (Department: 2863)
Data processing: measuring, calibrating, or testing
Measurement system in a specific environment
Chemical analysis
Reexamination Certificate
active
06675104
ABSTRACT:
FIELD OF THE INVENTION
Embodiments of the invention relate to methods for analyzing mass spectra.
BACKGROUND OF THE INVENTION
Recent advances in genomics research have led to the identification of numerous genes associated with various diseases. However, while genomics research can identify genes associated with a genetic predisposition to disease, there is still a need to characterize and identify markers such as proteins. A “marker” typically refers to a polypeptide or some other molecule that differentiates one biological status from another. Proteins and other markers are important factors in disease states. For example, proteins can vary in association with changes in biological states such as disease. They can also signal cellular responses to disease, toxicity, or other stimuli. When disease strikes, some proteins become dormant, while others become active. Prostate Specific Antigen (PSA), for example, is a circulating serum protein that, when elevated, correlates with prostate cancer. If the changes in protein levels could be rapidly detected, physicians could diagnose diseases early and improve treatments.
Identifying novel markers is one of the earliest and most difficult steps in the diagnostics and drug discovery processes. One way to discover if substances are markers for a disease is by determining if they are “differentially expressed” in biological samples from patients exhibiting the disease as compared to samples from patients not having the disease. For example, FIG.
1
(
a
) shows one graph
100
of a plurality of overlaid mass spectra of samples from a group of 18 diseased patients. The diseased patients could have, for example, prostate cancer. Another graph
102
is shown in FIG.
1
(
b
) and illustrates a plurality of overlaid mass spectra of samples from a group of 18 normal patients. In each of the graphs
100
,
102
, signal intensity is plotted as a function of mass-to-charge ratio. The intensities of the signals shown in the graphs
100
,
102
are proportional to the concentrations of markers having a molecular weight related to the mass-to-charge ratio A in the samples. As shown in the graphs
100
,
102
, at the mass-to-charge ratio A, a number of signals are present in both pluralities of mass spectra. The signals include peaks that represent potential markers having molecular weights related to the mass-to-charge ratio A.
When the signals in the graphs
100
,
102
are viewed collectively, it is apparent that the average intensity of the signals at the mass-to-charge ratio A is higher in the samples from diseased patients than the samples from the normal patients. The marker at the mass-to-charge ratio A is said to be “differentially expressed” in diseased patients, because the concentration of this marker is, on average, greater in samples from diseased patients than in samples from normal patients.
In view of the data shown in FIGS.
1
(
a
) and
1
(
b
), it can be generally concluded that the samples from diseased patients have a greater concentration of the marker with the mass-to-charge ratio A than the samples from normal patients. Since the concentration of the marker is generally greater in samples from diseased patients than in the normal samples, the marker can also be characterized as being “up-regulated” for the disease. If the concentration of the marker was generally less in the samples from diseased patients than in the samples from normal patients, the protein could be characterized as being “down-regulated”.
Once markers are discovered, they can be used as diagnostic tools. For example, with reference to the example described above, an unknown sample from a test patient may be analyzed using a mass spectrometer and a mass spectrum can be generated. The mass spectrum can be analyzed and the intensity of a signal at the mass-to-charge ratio A can be determined in the test patient's mass spectrum. The signal intensity can be compared to the average signal intensities at the mass-to-charge ratio A for diseased patients and normal patients. A prediction can then be made as to whether the unknown sample indicates that the test patient has or will develop cancer. For example, if the signal intensity at the mass-to-charge ratio A in the unknown sample is much closer to the average signal intensity at the mass-to-charge ratio A for the diseased patient spectra than for the normal patient spectra, then a prediction can be made that the test patient is more likely than not to develop or have the disease.
While the described differential expression analysis is useful, many improvements could be made. For instance, analyzing the amount of a single marker such as PSA in a patient's biological sample is many times not sufficiently reliable to monitor disease processes. PSA is considered to be one of the best prostate cancer markers presently available. However, it does not always correctly differentiate benign from malignant prostate disease. While the concentration of a marker such as PSA in a biological sample provides some ability to predict whether a test patient has a disease, an analytical method with a greater degree of reliability is desirable.
Also, when a large number of mass spectra of a large number of biological samples are analyzed, it is not readily apparent which signals represent markers that might differentiate between a diseased state and a non-diseased state. A typical mass spectrum of a biological sample has numerous potential marker signals (e.g., greater than 200) and a significant amount of noise. This can make the identification of potentially significant signals and the identification of average signal differentials difficult. Consequently, it is difficult to identify and quantify potential markers. Unless the potential markers exhibit strong up-regulation or strong down-regulation, the average signal differential between samples from diseased patients and samples from normal patients may not be easily discernable. For example, it is often difficult to visually determine that a cluster of signals at a given mass value in one group of mass spectra has higher or lower average signal intensity than a cluster of signals from another group of mass spectra. In addition, many potentially significant signals may have low intensity values. The noise in the spectra may obscure many of these potentially significant signals. The signals may go undiscovered and may be inadvertently omitted from a differential expression analysis.
It would be desirable to have better ways to analyze mass spectra. For example, it would be desirable to provide for a more accurate method for discovering potentially useful markers. It would also be desirable to provide an improved classification model that can be used to predict whether an unknown sample is associated or is not associated with a particular biological status.
Embodiments of the invention address these and other problems.
SUMMARY OF THE INVENTION
Embodiments of the invention relate to methods for analyzing mass spectra. In embodiments of the invention, a digital computer forms a classification model that can be used to differentiate classes of samples associated with different biological statuses. The classification model can be used as a diagnostic tool for prediction. It may also be used to identify potential markers associated with a biological status. In addition, the classification model can be formed using a process such as, for example, a recursive partitioning process.
One embodiment of the invention is directed to a method that analyzes mass spectra using a digital computer. The method comprises: entering into a digital computer a data set obtained from mass spectra from a plurality of samples, wherein each sample is, or is to be assigned to a class within a class set comprising two or more classes, each class characterized by a different biological status, and wherein each mass spectrum comprises data representing signal strength as a function of mass-to-charge ratio or a value derived from mass-to-charge ratio; and b) forming a classification model which discrimin
Braginsky Leonid
Fung Eric T.
Gavin Edward J.
Paulse Chris D.
Rich William E.
Barlow John
Ciphergen Biosystems Inc.
Lau Tung S
Townsend and Townsend / and Crew LLP
LandOfFree
Method for analyzing mass spectra does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for analyzing mass spectra, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for analyzing mass spectra will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3191191