System and method for identifying behavioral signatures
Psychopharmacological properties of new therapeutic drugs and highly heritable behavior patterns of test subjects are identified based on analysis of monitored exploratory movement to identify behavioral signatures. A test subject in a pen is allowed to explore for a period of time, after injecting it with a candidate drug or control vehicle. The test subject's movement is monitored and its locations stored. The locations are analyzed to separate them into behavioral patterns that are defined based on combinations of behavioral feature. Relative frequencies of performing each behavioral pattern are determined. In each pattern, differences between the relative frequencies in the candidate drug and control groups are tested, and only patterns in which this difference is highly significant are retained. The number of behavioral patterns further is reduced based on the relative frequencies and the correlation of behavioral patterns to one another, with the cells left over corresponding to a set of endpoints that identify a behavioral signature of the effect of the drug.
1. A computer-implemented method for identifying behavioral signatures for drug discovery, said method comprising:
recording tracking coordinates for each of a plurality of animals in an open-field test at a fixed rate for a period of time in a walled arena;
quantifying, by a computer, said tracking coordinates as behavioral data points of n dimensions, said n dimensions corresponding to n behavioral feature variables,
wherein said n behavioral feature variables include: time from a beginning of said period, momentary distance from a wall, momentary speed, momentary acceleration, momentary jerk of movement, momentary movement direction relative to said wall, and momentary path curvature;
for each of said n dimensions, dividing, by said computer, said behavioral data points into a plurality of intervals, such that each of said plurality of intervals contains an approximately equal number of behavioral data points;
defining, by said computer, a number of behavioral feature spaces of from 1 to n dimensions based on all possible combinations of said n dimensions, each of said number of behavioral feature spaces being divided into cells of said from 1 to n dimensions corresponding to each of said plurality of intervals for each of said from 1 to n dimensions;
dividing, by said computer, a number of behavioral data points within each cell by a total number of behavioral points for said open-field test to determine a relative frequency of said each cell;
analyzing, by said computer, a relative frequency of said each cell having a mean number of behavioral data points per test subject indicating said test subject spent on average at least one cumulative second performing a pattern of one or more of said n behavioral feature variables corresponding to said each cell; and
eliminating, by said computer, cells showing high cross-cell correlations by storing a cell having a maximal p-value and discarding other cells having a cross-correlation value greater than R 2 =0.4, and iteratively storing a remaining cell having a next highest maximal p-value and discarding other remaining cells having a subsequent cross-correlation value greater than R 2 =0.4, until there are no remaining cells, to produce most significant cells, which correspond to different patterns of movement,
wherein frequencies of said different patterns of movement by said plurality of animals constitute behavioral signatures for said drug discovery.
2. The method recited in claim 1 , further comprising:
dividing, by said computer, said behavioral data points into a progression mode or a lingering mode; and
subsequently using, by said computer, only said behavioral data points of said progression mode to produce said behavioral signatures.
3. The method recited in claim 1 , further comprising performing, by said computer, a logit transformation on said relative frequency of said cells.
4. The method recited in claim 1 , further comprising, analyzing, by said computer, differences between relative frequencies of said cells by statistical analysis of variance (ANOVA).
5. The method recited in claim 4 , further comprising applying, by said computer, a Bonferroni criterion, determined by 0.05 divided by a total number of said cells, to p-values of said cells to correct for multiple comparisons of said cells.
6. A computer system for identifying behavioral signatures for drug discovery, said system comprising:
a video camera for recording tracking coordinates for each of a plurality of animals in an open-field test at a fixed rate for a period of time in a walled arena;
a memory that stores said tracking coordinates; and
a processor configured to:
quantify said tracking coordinates as behavioral data points of n dimensions, said n dimensions corresponding to n behavioral feature variables,
wherein said n behavioral feature variables include: time from a beginning of said period, momentary distance from a wall, momentary speed, momentary acceleration, momentary jerk of movement, momentary movement direction relative to said wall, and momentary path curvature;
for each of said n dimensions, divide said behavioral data points into a plurality of intervals, such that each of said plurality of intervals contains an approximately equal number of behavioral data points;
define a number of behavioral feature spaces of from 1 to n dimensions based on all possible combinations of said n dimensions, each of said number of behavioral feature spaces being divided into cells of said from 1 to n dimensions corresponding to each of said plurality of intervals for each of said from 1 to n dimensions;
divide a number of behavioral data points within each cell by a total number of behavioral points for said open-field test to determine a relative frequency of said each cell;
analyze, by said computer, a relative frequency of said each cell having a mean number of behavioral data points per test subject indicating said test subject spent on average at least one cumulative second performing a pattern of one or more of said n behavioral feature variables corresponding to said each cell; and
eliminate cells showing high cross-cell correlations by storing a cell having a maximal p-value and discarding other cells having a cross-correlation value greater than R 2 =0.4, and iteratively storing a remaining cell having a next highest maximal p-value and discarding other subsequent remaining cells having a subsequent cross-correlation value greater than R 2 =0.4, until there are no subsequently remaining cells, to produce most significant cells, which correspond to different patterns of movement,
wherein frequencies of said different patterns of movement by said plurality of animals constitute behavioral signatures for said drug discovery.
7. The processor of the system recited in claim 6 , being further configured to:
divide said behavioral data points into a progression mode or a lingering mode; and
subsequently use only said behavioral data points of said progression mode to produce said behavioral signatures.
8. The processor of said system recited in claim 6 , being further configured to perform a logit transformation on said relative frequency of said cells.
9. The processor of said system recited in claim 6 , being further configured to analyze differences between relative frequencies of said cells by statistical analysis of variance (ANOVA).
10. The processor of said system recited in claim 9 being further configured to apply a Bonferroni criterion, determined by 0.05 divided by a total number of said cells, to p-values of said cells to correct for multiple comparisons of said cells.
11. A computer program storage medium readable by computer, tangibly embodying a program of instructions executable by said computer to perform a method for identifying a behavioral signature for drug discovery, said method comprising:
recording tracking coordinates for each of a plurality of animals in an open-field test at a fixed rate for a period of time in a walled arena;
quantifying said tracking coordinates as behavioral data points of n dimensions, said n dimensions corresponding to n behavioral feature variables,
wherein said n behavioral feature variables include: time from a beginning of said period, momentary distance from a wall, momentary speed, momentary acceleration, momentary jerk of movement, momentary movement direction relative to said wall, and momentary path curvature;
for each of said n dimensions, dividing said behavioral data points into a plurality of intervals, such that each of said plurality of intervals contains an approximately equal number of behavioral data points;
defining a number of behavioral feature spaces of from 1 to n dimensions based on all possible combinations of said n dimensions, each of said number of behavioral feature spaces being divided into cells of said from 1 to n dimensions corresponding to each of said plurality of intervals for each of said from 1 to n dimensions;
dividing a number of behavioral data points within each cell by a total number of behavioral points for said open-field test to determine a relative frequency of said each cell;
analyzing a relative frequency of said each cell having a mean number of behavioral data points per test subject indicating said test subject spent on average at least one cumulative second performing a pattern of one or more of said n behavioral feature variables corresponding to said each cell; and
eliminating cells showing high cross-cell correlations by storing a cell having a maximal p-value and discarding other cells having a cross-correlation value greater than R 2 =0.4, and iteratively storing a remaining cell having a next highest maximal p-value and discarding other remaining cells having a subsequent cross-correlation value greater than R 2 =0.4, until there are no remaining cells, to produce most significant cells, which correspond to different patterns of movement,
wherein frequencies of said different patterns of movement by said plurality of animals constitute behavioral signatures for said drug discovery.
12. The method of said computer program storage medium recited in claim 11 , further comprising:
dividing said behavioral data points into a progression mode or a lingering mode; and
subsequently using only said behavioral data points of said progression mode to produce said behavioral signatures.
13. The method of said computer program storage medium recited in claim 11 , further comprising performing a logit transformation on said relative frequency of said cells.
14. The method of said computer program storage medium recited in claim 11 , further comprising analyzing differences between relative frequencies of said cells by statistical analysis of variance (ANOVA).
15. The method of said computer program storage medium recited in claim 14 , further comprising applying a Bonferroni criterion, determined by 0.05 divided by a total number of said cells, to p-values of said cells to correct for multiple comparisons of said cells.