Limits...
The Marker State Space (MSS) method for classifying clinical samples.

Fallon BP, Curnutte B, Maupin KA, Partyka K, Choi S, Brand RE, Langmead CJ, Tembe W, Haab BB - PLoS ONE (2013)

Bottom Line: Marker State Space (MSS) defines "marker states" based on all possible patterns of high and low values among a panel of markers.Each marker state is defined as either a case state or a control state, and a sample is classified as case or control based on the state it occupies.MSS provides a straightforward approach for modeling highly divergent subclasses of patients, which may be adaptable for diverse applications.

View Article: PubMed Central - PubMed

Affiliation: Laboratory of Cancer Immunodiagnostics, Van Andel Institute, Grand Rapids, Michigan, USA.

ABSTRACT
The development of accurate clinical biomarkers has been challenging in part due to the diversity between patients and diseases. One approach to account for the diversity is to use multiple markers to classify patients, based on the concept that each individual marker contributes information from its respective subclass of patients. Here we present a new strategy for developing biomarker panels that accounts for completely distinct patient subclasses. Marker State Space (MSS) defines "marker states" based on all possible patterns of high and low values among a panel of markers. Each marker state is defined as either a case state or a control state, and a sample is classified as case or control based on the state it occupies. MSS was used to define multi-marker panels that were robust in cross validation and training-set/test-set analyses and that yielded similar classification accuracy to several other classification algorithms. A three-marker panel for discriminating pancreatic cancer patients from control subjects revealed subclasses of patients based on distinct marker states. MSS provides a straightforward approach for modeling highly divergent subclasses of patients, which may be adaptable for diverse applications.

Show MeSH

Related in: MedlinePlus

Determining optimal thresholds for a two-marker panel.(A) Scanning thresholds. Three different thresholds are depicted for Marker 1 (left) and Marker 2 (right), with the resulting conversion to 1s and 0s for each threshold, followed by the sensitivities and specificities for each marker at each threshold. (B) Determining the best combination of thresholds. All possible combinations of thresholds were assembled for the two-marker panel, resulting in nine combinations. Based on the results from panel A, the numbers of cancer and non-cancer samples that occupy each state were determined for each combination, from which the sensitivity and specificity could be calculated for each combination. The combination of thresholds giving the best performance (in this case threshold 2 for Marker 1 and threshold 2 for Marker 2) is selected.
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC3672150&req=5

pone-0065905-g002: Determining optimal thresholds for a two-marker panel.(A) Scanning thresholds. Three different thresholds are depicted for Marker 1 (left) and Marker 2 (right), with the resulting conversion to 1s and 0s for each threshold, followed by the sensitivities and specificities for each marker at each threshold. (B) Determining the best combination of thresholds. All possible combinations of thresholds were assembled for the two-marker panel, resulting in nine combinations. Based on the results from panel A, the numbers of cancer and non-cancer samples that occupy each state were determined for each combination, from which the sensitivity and specificity could be calculated for each combination. The combination of thresholds giving the best performance (in this case threshold 2 for Marker 1 and threshold 2 for Marker 2) is selected.

Mentions: The discovery of biomarker panels based on this classification system requires a method for selecting the members of the marker panel, the thresholds for each marker, and the state rules (the designation of which states are cases and which states are controls). These three factors, the markers, the thresholds, and the state rules, are related to each other, so that changes in one might affect the optimal values for the other two. An approach to selecting the thresholds and state rules that best discriminate two groups of samples is illustrated in Figure 2 for two markers. For each individual marker, several test thresholds are applied to convert the data to 1s and 0s (Fig. 2A). To determine which thresholds work best together between the markers, all nine possible combinations could be examined (Fig. 2B). For each of these combinations, we can assign certain states to indicate cases and other states to indicate controls. A simple approach to making those assignments is to count how many case and control samples populate each state, and then make the assignment accordingly (Fig. 2B). For example, if state 0,1 is populated by six control samples and only two case samples, the state would be assigned to indicate controls. Once the assignment is made for each state, all samples in each state are classified according to the assignments. A sensitivity and specificity can be calculated based on how many case and control samples were correctly classified.


The Marker State Space (MSS) method for classifying clinical samples.

Fallon BP, Curnutte B, Maupin KA, Partyka K, Choi S, Brand RE, Langmead CJ, Tembe W, Haab BB - PLoS ONE (2013)

Determining optimal thresholds for a two-marker panel.(A) Scanning thresholds. Three different thresholds are depicted for Marker 1 (left) and Marker 2 (right), with the resulting conversion to 1s and 0s for each threshold, followed by the sensitivities and specificities for each marker at each threshold. (B) Determining the best combination of thresholds. All possible combinations of thresholds were assembled for the two-marker panel, resulting in nine combinations. Based on the results from panel A, the numbers of cancer and non-cancer samples that occupy each state were determined for each combination, from which the sensitivity and specificity could be calculated for each combination. The combination of thresholds giving the best performance (in this case threshold 2 for Marker 1 and threshold 2 for Marker 2) is selected.
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC3672150&req=5

pone-0065905-g002: Determining optimal thresholds for a two-marker panel.(A) Scanning thresholds. Three different thresholds are depicted for Marker 1 (left) and Marker 2 (right), with the resulting conversion to 1s and 0s for each threshold, followed by the sensitivities and specificities for each marker at each threshold. (B) Determining the best combination of thresholds. All possible combinations of thresholds were assembled for the two-marker panel, resulting in nine combinations. Based on the results from panel A, the numbers of cancer and non-cancer samples that occupy each state were determined for each combination, from which the sensitivity and specificity could be calculated for each combination. The combination of thresholds giving the best performance (in this case threshold 2 for Marker 1 and threshold 2 for Marker 2) is selected.
Mentions: The discovery of biomarker panels based on this classification system requires a method for selecting the members of the marker panel, the thresholds for each marker, and the state rules (the designation of which states are cases and which states are controls). These three factors, the markers, the thresholds, and the state rules, are related to each other, so that changes in one might affect the optimal values for the other two. An approach to selecting the thresholds and state rules that best discriminate two groups of samples is illustrated in Figure 2 for two markers. For each individual marker, several test thresholds are applied to convert the data to 1s and 0s (Fig. 2A). To determine which thresholds work best together between the markers, all nine possible combinations could be examined (Fig. 2B). For each of these combinations, we can assign certain states to indicate cases and other states to indicate controls. A simple approach to making those assignments is to count how many case and control samples populate each state, and then make the assignment accordingly (Fig. 2B). For example, if state 0,1 is populated by six control samples and only two case samples, the state would be assigned to indicate controls. Once the assignment is made for each state, all samples in each state are classified according to the assignments. A sensitivity and specificity can be calculated based on how many case and control samples were correctly classified.

Bottom Line: Marker State Space (MSS) defines "marker states" based on all possible patterns of high and low values among a panel of markers.Each marker state is defined as either a case state or a control state, and a sample is classified as case or control based on the state it occupies.MSS provides a straightforward approach for modeling highly divergent subclasses of patients, which may be adaptable for diverse applications.

View Article: PubMed Central - PubMed

Affiliation: Laboratory of Cancer Immunodiagnostics, Van Andel Institute, Grand Rapids, Michigan, USA.

ABSTRACT
The development of accurate clinical biomarkers has been challenging in part due to the diversity between patients and diseases. One approach to account for the diversity is to use multiple markers to classify patients, based on the concept that each individual marker contributes information from its respective subclass of patients. Here we present a new strategy for developing biomarker panels that accounts for completely distinct patient subclasses. Marker State Space (MSS) defines "marker states" based on all possible patterns of high and low values among a panel of markers. Each marker state is defined as either a case state or a control state, and a sample is classified as case or control based on the state it occupies. MSS was used to define multi-marker panels that were robust in cross validation and training-set/test-set analyses and that yielded similar classification accuracy to several other classification algorithms. A three-marker panel for discriminating pancreatic cancer patients from control subjects revealed subclasses of patients based on distinct marker states. MSS provides a straightforward approach for modeling highly divergent subclasses of patients, which may be adaptable for diverse applications.

Show MeSH
Related in: MedlinePlus