Authors (@slack): Malak Abdelfattah Soula (@Malak)
Machine learning(ML), is a branch of artificial intelligence and computer science. When no equation describes a problem, it learns from data. By providing the dataset to the ML algorithm, it begins to build the model that it will be used for rather than to conclude the equation that will predict the output. This essay begins by defining the ML concept, especially in bioinformatics. It then explores the application of ML in biology. Subsequently, it describes the phases of ML modeling.
Machine learning is the ability to use computers to solve problems, by learning from historical data to predict the desired output that will help enhance the decision making. In bioinformatics, machine learning can help by building models that can be used to detect patterns within biological inputs like genomes, healthcare data, etc. ML trains from these biological data and makes predictions ML techniques are applied in two main biological domains:(1) omics science and (2) systems biology. It can be applied to analyzing biological data from various areas like predicting the gene locations and their biological roles, predicting protein or molecular functions, and studying the interactions between them Also, predicting diseases from molecular samples and more and more.
Figure1: shows the application of machine learning in biology.
ML has two major phases which are training and testing the model. That is used to answer biological questions. In the training phase, the ML model begins to train in the training data the model has, it is important so that the model can make predictions. Followed by a testing phase where the testing data is used to evaluate the performance of the model and make predictions.
Figure 2 : shows the two main phases of ML.
ML has main categories are listed as follows: Supervised learning: this class mainly depends on predefined input labels, and the model begins to find the correlation between the input data and target output which is the target. This class has two types :(1) classification and the common classification algorithms are Decision trees, Random forests, Support vector machines, Neural Networks, Nearest Neighbor, and Naïve Bayes. (2) regression and its algorithm include linear or multivariate regression logistic regression with lasso, ridge, or elastic-net penalties. Unsupervised learning: where the machine trains with input data or labels only, and the output is unknown and tries to detect patterns. The algorithm used in unsupervised learning is the clustering algorithm, where the data points with the same properties are classified together to describe hidden structures from unlabeled data. Semi-supervised learning: here the dataset is mixed between labeled and unlabeled data, also known as inductive learning.
Figure 3: shows the different types of ML.
In conclusion, ML techniques and tools are useful in the bioinformatics era. Begin by training the model to make predictions and make investigations. These powerful technologies have opened up new possibilities for understanding and manipulating biological systems as they grow exponentially with time.
1- Larrañaga, P., Calvo, B., Santana, R., Bielza, C., Galdiano, J., Inza, I., Lozano, J.A., Armañanzas, R., Santafé, G., Pérez, A. and Robles, V. (2006). Machine learning in bioinformatics. Briefings in Bioinformatics, 7(1), pp.86–112. doi:https://doi.org/10.1093/bib/bbk007.
2- Semanticscholar.org. (2021). [PDF] Incorporating Machine Learning into Established Bioinformatics Frameworks | Semantic Scholar. [online] Available at: https://www.semanticscholar.org/reader/779ae53b1d60b34bdf88a6b23af82e62854197c3 [Accessed 3 Sep. 2024].