Classification of caenorhabditis elegans genotypes using locomotory behavioural patterns

Abstract

Understanding the neural basis of decision making is a major challenge in many disciplines. One way to study the behaviour of a model organism is through their movement patterns in a well-designed environment. The Caenorhabditis elegans (C .elegans) is an example of a model organism used for many biological investigation. C .elegans often moves from one location to another especially in search of food or due to environmental changes or threat. As C .elegans moves, the changes that occurs in their body shapes provide information that could aid in understanding the worm’s locomotory activity. In this work, we have built a computational model that automates the retrieval and processing of features from the skeleton body shapes of the worms in a video sequence. These measurements are used to quantify and classify between a wild-type worm (AQ2947) and different classes of mutant worms ( OW939, OW940, OW949, OW953 and OW956 ). In the previous works, the worm genotypes classification was done using features extracted from the worm’s posture phenotypes as input to classifiers such as random forest, decision tree and a deep classification model. However, it is not clear if these methods provide a compact representation of the worm’s movement activity. Here, we used both supervised and unsupervised dimensionality reduction methods such as Principal Component Analysis (PCA), Non-negative Factorisation Matrix (NMF), Linear Discriminant Analysis (LDA), Kernel Principal Component Analysis (KPCA), and Kernel Fisher Linear Discriminant Analysis (KLDA) to extract low-dimensional representations of the skeleton angle data derived from each image frame in a movie. These low-dimensional features serves as input to standard machine learning algorithms such as k- nearest neighbour (kNN), random forest (RF), and support vectors machine (SVM), and a deep classification model. The outcome of this investigation shows that postural features of the worms retrieved over a period of time can be use to classify their genotypes.

More information

Identifiers

ORCID for Samuel, Apigi Ikirigo: orcid.org/0000-0002-6065-4765

Catalogue record

Export record