Limits...
Hidden Markov model analysis of maternal behavior patterns in inbred and reciprocal hybrid mice.

Carola V, Mirabeau O, Gross CT - PLoS ONE (2011)

Bottom Line: Individual variation in maternal care in mammals shows a significant heritable component, with the maternal behavior of daughters resembling that of their mothers.For the HMM analysis we defined seven states: arched-backed nursing, blanket nursing, licking/grooming pups, grooming, activity, eating, and sleeping.Differences in these patterns observed in the experimental groups (inbred and hybrid females) were detected only after the application of HMM analysis whereas classical statistical methods and analyses were not able to highlight them.

View Article: PubMed Central - PubMed

Affiliation: Mouse Biology Unit, European Molecular Biology Laboratory, Monterotondo, Italy. carola@embl.it

ABSTRACT
Individual variation in maternal care in mammals shows a significant heritable component, with the maternal behavior of daughters resembling that of their mothers. In laboratory mice, genetically distinct inbred strains show stable differences in maternal care during the first postnatal week. Moreover, cross fostering and reciprocal breeding studies demonstrate that differences in maternal care between inbred strains persist in the absence of genetic differences, demonstrating a non-genetic or epigenetic contribution to maternal behavior. In this study we applied a mathematical tool, called hidden Markov model (HMM), to analyze the behavior of female mice in the presence of their young. The frequency of several maternal behaviors in mice has been previously described, including nursing/grooming pups and tending to the nest. However, the ordering, clustering, and transitions between these behaviors have not been systematically described and thus a global description of maternal behavior is lacking. Here we used HMM to describe maternal behavior patterns in two genetically distinct mouse strains, C57BL/6 and BALB/c, and their genetically identical reciprocal hybrid female offspring. HMM analysis is a powerful tool to identify patterns of events that cluster in time and to determine transitions between these clusters, or hidden states. For the HMM analysis we defined seven states: arched-backed nursing, blanket nursing, licking/grooming pups, grooming, activity, eating, and sleeping. By quantifying the frequency, duration, composition, and transition probabilities of these states we were able to describe the pattern of maternal behavior in mouse and identify aspects of these patterns that are under genetic and nongenetic inheritance. Differences in these patterns observed in the experimental groups (inbred and hybrid females) were detected only after the application of HMM analysis whereas classical statistical methods and analyses were not able to highlight them.

Show MeSH

Related in: MedlinePlus

Hidden Markov model labeling of maternal behavior.An HMM is characterized by a state transition probability matrix A, an observation probability matrix B, and a state probability matrix π. The Baum-Welch algorithm is used to calculate a final HMM from an initial HMM (with user-defined, estimated probabilities) by maximizing the likelihood of emitting the observed behavioral sequence data. The final HMM is then used to label the behavioral sequence of each subject by application of the Viterbi algorithm. A statistical assessment of the frequency, duration, composition, and transitions between these labels can then be used to document strain differences in behavior.
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC3050935&req=5

pone-0014753-g002: Hidden Markov model labeling of maternal behavior.An HMM is characterized by a state transition probability matrix A, an observation probability matrix B, and a state probability matrix π. The Baum-Welch algorithm is used to calculate a final HMM from an initial HMM (with user-defined, estimated probabilities) by maximizing the likelihood of emitting the observed behavioral sequence data. The final HMM is then used to label the behavioral sequence of each subject by application of the Viterbi algorithm. A statistical assessment of the frequency, duration, composition, and transitions between these labels can then be used to document strain differences in behavior.

Mentions: Based on our previous observations [10] we specified seven HMM states: blanket nursing (BLN), arched-back nursing (ABN), licking/grooming pups (LG), grooming (GRO), activity (ACT), eating (EAT), and sleeping (SLP). Initial conditions for the HMM were chosen consistent with our previous data (see Table S1, S2, and S3) [10] and a single final HMM, named MatHMM, was derived by application of the Baum-Welch algorithm to maternal behavior data from mothers of all four strains (Ntotal = 127) during the first postnatal week (Figure 2).


Hidden Markov model analysis of maternal behavior patterns in inbred and reciprocal hybrid mice.

Carola V, Mirabeau O, Gross CT - PLoS ONE (2011)

Hidden Markov model labeling of maternal behavior.An HMM is characterized by a state transition probability matrix A, an observation probability matrix B, and a state probability matrix π. The Baum-Welch algorithm is used to calculate a final HMM from an initial HMM (with user-defined, estimated probabilities) by maximizing the likelihood of emitting the observed behavioral sequence data. The final HMM is then used to label the behavioral sequence of each subject by application of the Viterbi algorithm. A statistical assessment of the frequency, duration, composition, and transitions between these labels can then be used to document strain differences in behavior.
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC3050935&req=5

pone-0014753-g002: Hidden Markov model labeling of maternal behavior.An HMM is characterized by a state transition probability matrix A, an observation probability matrix B, and a state probability matrix π. The Baum-Welch algorithm is used to calculate a final HMM from an initial HMM (with user-defined, estimated probabilities) by maximizing the likelihood of emitting the observed behavioral sequence data. The final HMM is then used to label the behavioral sequence of each subject by application of the Viterbi algorithm. A statistical assessment of the frequency, duration, composition, and transitions between these labels can then be used to document strain differences in behavior.
Mentions: Based on our previous observations [10] we specified seven HMM states: blanket nursing (BLN), arched-back nursing (ABN), licking/grooming pups (LG), grooming (GRO), activity (ACT), eating (EAT), and sleeping (SLP). Initial conditions for the HMM were chosen consistent with our previous data (see Table S1, S2, and S3) [10] and a single final HMM, named MatHMM, was derived by application of the Baum-Welch algorithm to maternal behavior data from mothers of all four strains (Ntotal = 127) during the first postnatal week (Figure 2).

Bottom Line: Individual variation in maternal care in mammals shows a significant heritable component, with the maternal behavior of daughters resembling that of their mothers.For the HMM analysis we defined seven states: arched-backed nursing, blanket nursing, licking/grooming pups, grooming, activity, eating, and sleeping.Differences in these patterns observed in the experimental groups (inbred and hybrid females) were detected only after the application of HMM analysis whereas classical statistical methods and analyses were not able to highlight them.

View Article: PubMed Central - PubMed

Affiliation: Mouse Biology Unit, European Molecular Biology Laboratory, Monterotondo, Italy. carola@embl.it

ABSTRACT
Individual variation in maternal care in mammals shows a significant heritable component, with the maternal behavior of daughters resembling that of their mothers. In laboratory mice, genetically distinct inbred strains show stable differences in maternal care during the first postnatal week. Moreover, cross fostering and reciprocal breeding studies demonstrate that differences in maternal care between inbred strains persist in the absence of genetic differences, demonstrating a non-genetic or epigenetic contribution to maternal behavior. In this study we applied a mathematical tool, called hidden Markov model (HMM), to analyze the behavior of female mice in the presence of their young. The frequency of several maternal behaviors in mice has been previously described, including nursing/grooming pups and tending to the nest. However, the ordering, clustering, and transitions between these behaviors have not been systematically described and thus a global description of maternal behavior is lacking. Here we used HMM to describe maternal behavior patterns in two genetically distinct mouse strains, C57BL/6 and BALB/c, and their genetically identical reciprocal hybrid female offspring. HMM analysis is a powerful tool to identify patterns of events that cluster in time and to determine transitions between these clusters, or hidden states. For the HMM analysis we defined seven states: arched-backed nursing, blanket nursing, licking/grooming pups, grooming, activity, eating, and sleeping. By quantifying the frequency, duration, composition, and transition probabilities of these states we were able to describe the pattern of maternal behavior in mouse and identify aspects of these patterns that are under genetic and nongenetic inheritance. Differences in these patterns observed in the experimental groups (inbred and hybrid females) were detected only after the application of HMM analysis whereas classical statistical methods and analyses were not able to highlight them.

Show MeSH
Related in: MedlinePlus