Week 1-4: Alignments and Machine Learning

Question 1

Define: genome

Answer 1

All of the genetic information in an organism, can be either DNA or RNA. Usually refers to the genetic sequence.

Question 2

Define: transcriptome

Answer 2

All of the transcribed RNA in an organism. Not all DNA is transcribed

Question 3

Define: proteome

Answer 3

The entire complement of proteins that is or can be expressed by a cell, tissue, or organism

Question 4

Describe: homology (genes/proteins)

Answer 4

Proteins or genes are defined as homologous if they can be said to have shared an ancestor. Genes or proteins are either homologs or they are not; there is no such thing as percent homology, however % identity of sequences can be identified.

Question 5

Define: Homology

Answer 5

Descent from a common ancestor

Question 6

Define: Orthology

Answer 6

Descent from a speciation event

Question 7

Define: Parology

Answer 7

Descent from a duplication event

Question 8

Define: Xenology

Answer 8

Descent from a horizontal transfer event

Question 9

When can homology be determined as ‘real’?

Answer 9

If in a pairwise alignment >25% identical amino acids as proteins will have a similar folding pattern. 18-25% identity - possible homology. <18% - cannot be determined from alignment

Question 10

When are DNA alignments appropriate for use (over protein alignments)?

Answer 10

1. To confirm identity of cDNA
2. To study non-coding DNA
3. To study DNA polymorphisms

Question 11

Define: Global alignment

Answer 11

Aligns entirety of two sequences, finding global (overall) similarity.

Question 12

Define: Local alignment

Answer 12

Looks for (local) regions of similarity

Question 13

PAM Matrices

Answer 13

Multiple alignments of related sequences (>85%), looking at substitutions. 
1 PAM (point accepted mutation) = one change per hundred aa's.

Question 14

Machine learning: Supervised Learning

Answer 14

• Training data are labelled e.g. malignant or benign

- Goal is to correctly predict the output value given an unlabelled data point

Question 15

Machine learning: Unsupervised Learning

Answer 15

• Input data are not labelled (output data not attached) - underlying class not important
• Algorithm must find patterns in data

Question 16

Machine learning: Reinforcement Learning

Answer 16

Algorithm learns through occasional ‘rewards’ or ‘punishments’, adjusting it’s behaviour based on this feedback

Question 17

Machine learning: Supervised Learning categories

Answer 17

• Classification: categorical labels e.g. skin cancer types (malignant, non-malignant, melanoma etc.)
• Regression: continuous labels e.g. predicting drug half life
• Ordinal regression: categorical labels that has some natural ordering e.g. predicting cancer stages (0-IV)

Question 18

Machine learning: Supervised learning: regression

Answer 18

• Predicting continuous output values: gene expression levels, drug half-life or patient life-expectancy (continuous range of output values).
• A function or algorithm that maps a set of known input features to a real valued output value
• Training of a regression algorithm often involves minimising a cost function

Question 19

Machine learning: Linear regression

Answer 19

• Assumes a linear relationship between continuous input variables and output value
• Can be problematic with highly dimensional data
• Aim is to find the ‘line of best fit’ through the data

Question 20

Machine learning: Cost function

Answer 20

a way of quantifying the difference between prediction and reality

Question 21

Machine learning: Support Vector Machine

Answer 21

A classification (supported learning) algorithm.

Question 22

Machine learning: Supervised learning - classification

Answer 22

• Function or algorithm that maps set of known input features to a categorical output value, i.e. predicting categorical output values using known input data values, e.g. cancer status, protein secondary structure, or immune recognition (epitope/non-epitope)

Question 23

Machine learning: Classification algorithms

Answer 23

Support vector machine and neural networks

Question 24

Machine learning: Support vector machine

Answer 24

• Classification (supervised learning) algorithm
• SVM maximises separation between classes
• Only the values of support vectors affect decision boundary
• If non-linear separation is needed (for data not linearly separable), can use Kernel function/trick

Question 25

Machine learning: Kernel trick

Answer 25

- Kernels allow us to use this mapping into higher dimensions without having to explicitly compute the transformation for each data point - Input vectors only appear once, as a dot product

Question 26

Machine learning: Dot product

Answer 26

- ‘Similarity measure’ between two input data points - If input data is far apart, dot product will be close to zero. - Can be replaced with any equation that measures similarity between two data points

Question 27

A Kernel function only needs to compute ____ between two input data points

Answer 27

similarity

Question 28

Machine learning: Neural Networks

Answer 28

Each node outputs a value based on a combination of input values Each node ‘learns’ which inputs are important Can have multiple hidden layers Often need large amounts of training data for algorithms to perform well

Question 29

Machine learning: hierarchical Clustering

Answer 29

A clustering (unsupervised learning) algorithm Important for discovering biologically relevant clusters/groups Input data isn't labeled

Question 30

Anatomy of a Machine Learning Problem

Answer 30

1. Data collection: More data is nearly always better, however most biological datasets are 'small' in the context of modern machine learning but despite relatively small dataset size (number of samples), can often have a large number of features (highly dimensional) e.g. gene expression data. Hence arises need for: 2. Feature extraction/selection: Conversion of biological observations into computer-friendly representations (features) in a process known as feature extraction. This can be driven by biological insight (if important contributing factors are known), followed by feature selection (reducing observations to informative features). 3. Model selection and training: Data is split into training, validation and test datasets. This prevents overfitting (good performance during training but poor generalisation of outcome). The model is trained on a training dataset. Performance is then assessed on a validation dataset and hyperparameters used to improve performance. The best model is then evaluated on a test dataset.

Question 31

If performance of a machine learning model is poor, what can be the contributing factors?

Answer 31

Poor algorithm choice. Insufficient data. Noisy dataset.

Question 32

Machine learning evaluation metrics - accuracy

Answer 32

What proportion of predictions are correct? | TP+TN/Total number of predictions

Question 33

Machine learning evaluation metrics - precision

Answer 33

What proportion of our predicted positives are true?(TP/TP+FP)

Question 34

Machine learning evaluation metrics - recall

Answer 34

What proportion of true positives did we correctly identify?(TP/TP+FN)

Question 35

Machine learning problem: Overfitting

Answer 35

excellent performance on the training dataset, but poor generalisation to unseen data A big problem when training data has many features, or few data points

Question 36

Machine learning problem: redundancy between test and training sets

Answer 36

Very similar data points in both test and training data, leading to overestimation of model performance

Question 37

Machine learning problem: Unbalanced classes within the dataset

Answer 37

Can be very difficult to train a model when one class dominates the training data

Question 38

Machine learning problem: Batch effect

Answer 38

Data collection is not properly randomised | Model won’t generalise to unseen data

Question 39

Define: Systematics

Answer 39

An attempt to understand the interrelationships of living things

Question 40

Taxonomy

Answer 40

The science of naming and classifying organisms (evolutionary theory not necessarily involved)

Question 41

Phylogenetics

Answer 41

The field of systematics that focuses on evolutionary relationships between organisms or genes/proteins (phylogeny).

Question 42

Define: Taxon

Answer 42

Any named group of organisms (evolutionary theory not necessarily involved)

Question 43

A group of taxa proposed to have a common ancestral origin is called

Answer 43

clade

Question 44

The base of a clade is called a ______

Answer 44

node

Question 45

Define: Topology

Answer 45

Order of branching in a phylogenetic tree

Question 46

Define: Cladistics

Answer 46

Classifying organisms based on revolutionary relatedness or shared characteristics

Question 47

True or false: | In a cladogram, branch lengths indicate quantity of evolutionary divergence.

Answer 47

False. In a cladogram, branch lengths are not significant. Only the topology (order of branching) matters.

Question 48

True or false: | In a phylogram, branch lengths indicate quantity of evolutionary divergence.

Answer 48

True.

Question 49

In a phylogenetic tree, clades show a common ancestor and _____

Answer 49

all of its descendants.

Question 50

True or false: | In a phylogenetic tree, terminal nodes beside each other are more closely related than tips further apart.

Answer 50

False.

Question 51

True or false: | Maximum Parsimony is based on sequence characters rather than distances

Answer 51

True

Question 52

Bootstrapping

Answer 52

A way of statistically validating a phylogenetic tree. Data is resampled (generally 1000 times) after being slightly perturbed and the number (or percentage) of times a node appears is given. If a node is present 700 times from 1000, around 95% probability it is in the correct position

Question 53

Molecular clock

Answer 53

Technique that uses mutation rate of biomolecules to deduce the time taken for two or more species to diverge

Question 54

Dynamic programming in sequence alignment - steps

Answer 54

1. Initialisation 2. Scoring the matrix 3. Traceback (to get the alignment)

Question 55

PAM250 refers to

Answer 55

250 substitutions per 100 amino acids

Question 56

BLOSUM62 is derived from

Answer 56

proteins that have no more than 62% identity

Question 57

BLASTP uses which default scoring matrix?

Answer 57

BLOSUM62

Question 58

E value

Answer 58

The 'expect' value - the probability that the search will show a match by chance

Question 59

Classification supervised learning can be used for

Answer 59

Skin cancer type (melanoma, non-malignant, basal-cell carcinoma etc.)

Question 60

Regression supervised learning can be used for

Answer 60

Predicting drug half-life

Question 61

Ordinal regression supervised learning can be used for

Answer 61

Predicting cancer stages (Stage I-IV)

Question 62

Linear regression can be problematic with _______ data.

Answer 62

highly dimensional