(Week 3) [T6] Precision Medicine Omics Data Analysis

Question 1

What can we learn from omics data in terms of genomics?

Answer 1

What are the genome variants related with disease or drug.

Question 2

What can we learn from omics data in terms of epigenomics?

Answer 2

Which genes are active/silenced. Modifications correlated with diseases or response to drugs.

Question 3

What can we learn from omics data in terms of metabolomics?

Answer 3

Which metabolites are present and how much.

Question 4

What can we learn from omics data in terms of interctomics?

Answer 4

How genes and RNA molecules interact with each other.

Question 5

What can we learn from omics data in terms of microbiomics?

Answer 5

Which microbes are present in our organism.

Question 6

Give an example of data processing technique, statistics and dimensionality reduction.

Answer 6

• Normalization; aggregation.
• Fisher’s exact test; pearson chi-square.
• PCA.

Question 7

Which are the 3 types of machine learning?

Answer 7

Unsupervised
- Clustering. Divide data in groups, in a way that intracluster distances are minimized and intercluster distances are maximized.
- Pattern mining and subspace clustering. Find meaningful correlations on a subset of the feature space.

Supervised
- Classification. Classification could be binary or multi-class.
- Regression. Learn a mapping to estimate the outcome of a new observation.

Semi-supervised have few labelled individuals due to label acquisition costs and uncertainty

Question 8

Which are the evaluation techniques for classification models?

Answer 8

• Train-test split. Learn model from train data and assess it against test data.
• Cross validation. Divide data into k subsets/folds and use 1 fold for test and remaining (k-1) subsets for train. Repeat for all k folds.
• Prevalence. Percentage of the population with the condition.
• F-measure = 2 Precision x Recall / (Precision + Recall).
• Confusion matrix.