Computational biology
Learn Python, biological data science, machine learning, systems biology and computational neuroscience through real research problems.
9 modules · 35 lessons · 45h · mastery threshold 80
Watch videos free — no sign-inLearn Python, biological data science, machine learning, systems biology and computational neuroscience through real research questions. Designed for ambitious high school and early undergraduate learners preparing for a mentored computational research project. By the end, you can analyze biological datasets, visualize evidence, evaluate simple models, interpret genomic and neural data, model dynamics, state limitations and formulate a defensible research question. Four applied labs and a submitted capstone lead toward instructor review and a future oral defense.
Module 1
Turn biological questions into samples, features and documented measurements.
Module 2
Use Python to make biological measurements traceable and reproducible.
Module 3
Describe variation, quantify uncertainty and make honest figures.
Module 4
Build and evaluate biological predictions without leakage or causal overclaiming.
Module 5
Represent sequences and expression matrices and interpret exploratory structure.
Module 6
Model changing biological systems, regulation and feedback.
Module 7
Combine signals responsibly and examine clinical validity and bias.
Module 8
Analyze spikes and EEG-style signals while respecting inference limits.
Module 9
Design, reproduce and defend a scoped computational research project.
Activation, inhibition and feedback
Which feedback can oppose a perturbation?
Alignment and gaps
Which method can represent an insertion between sequences?
Alignment and k-mers
Sequence similarity can suggest shared ancestry or function, but chance matches also occur.
Array shape and vectorization
An array stores measurements in a shape such as samples by features.
Baselines and prespecified analysis
Which item is a useful baseline for an imbalanced classifier?
Biological prediction tasks
A regression output is numerical, such as biomarker concentration; a classifier estimates a class or class probability.
Calibration and distribution shift
A 20% predicted risk should correspond to what under calibration?
Choosing plots for distributions
Which plot best reveals a skewed one-variable distribution?
Classification metrics and imbalance
A confusion matrix counts true positives, false positives, true negatives and false negatives.
Clustering versus causal interpretation
What should you check when clusters match collection date?
Collection batch confounding
Why record a collection batch?
Continuous targets and regression
A future blood concentration is continuous. Which task fits?
Control flow and reusable analysis
A condition encodes a scientific decision; a loop applies it to each sample; a function packages the rule for testing and reuse.
DataFrame filtering and aggregation
A DataFrame gives each row an observation and each column a measured or recorded variable.
Distribution and variability
Mean and median describe centers differently when outliers are present.
DNA base pairing
Which bases pair in DNA?
EEG and spectral analysis
EEG records scalp voltage influenced by many neural and non-neural sources.
EEG bands and artifacts
What is a common non-neural EEG artifact?
Excitation and inhibition
What can an inhibitory input do?
Experimental design and confounding
A treatment is an independent variable; an outcome is a dependent variable.
Expression matrix interpretation
In an expression matrix, samples are observations and genes are features.
External validation and calibration
A clinically useful model must predict an outcome available at the intended decision time.
Falsifiable research hypotheses
What makes a hypothesis falsifiable?
Functions and loop logic
What should change when applying one threshold to 100 samples?
Gene-expression matrix orientation
In a samples-by-genes matrix, which axis lists genes?
Generalization and data leakage
Training fits model parameters; validation guides choices; a held-out test estimates performance on unseen cases.
Genotype and phenotype
A DNA variant is a genotype measurement. Which is a phenotype?
Heterogeneity and patient stratification
Population averages can hide different responses across individuals.
Hill saturation and cooperativity
What happens to a saturating activation curve far above K?
Honest scientific visualization
Choose a plot to answer a question: histogram for distribution, scatter for relationship, box or violin for group spread.
Immutable raw data and negative results
Why keep raw data immutable?
Longitudinal patient data
Two visits from one patient are best treated as?
Missing assay values
A missing assay value is encoded as NaN. What should happen first?
Molecular information flow
DNA sequence is transcribed into RNA; coding RNA can be translated into protein.
Multimodal sample matching
What key should connect two modalities?
Multimodal systems medicine
Genomics, transcriptomics, proteomics, imaging and physiology measure different aspects of a system.
Networks and feedback
A biological network represents entities as nodes and influences as activating or inhibiting interactions.
Neural signaling
Membrane potential reflects electrical and chemical gradients across a neuron.
Neural-data classification
Turn a recording into samples such as patient-level sessions, compute features from training data and evaluate on held-out people.
Nonlinear gene regulation
Transcription factors can activate or repress expression.
NumPy shape and broadcasting
How many values result from mean(axis=1) on a 3 by 2 matrix?
Overfitting and interpretation
An overly flexible model can fit noise in training data and fail on new data.
P-values and effect size
A tiny p-value accompanies a tiny effect. What else should be reported?
PCA and exploratory clustering
PCA finds directions of large variation; a two-dimensional PCA plot is a projection that discards some information.
Person-level neural-data splitting
How should EEG windows from one person be partitioned?
Rate of change and steady state
What is the steady state of dx/dt=6-2x?
Rates, parameters and steady states
A differential equation describes how quickly a state changes.
Recognizing overfitting
What pattern suggests overfitting?
Regression versus classification
Predicting exact glucose concentration is which task?
Reproducible analysis and limits
Keep raw data immutable and save derived data with documented transformations.
Research analysis plan
Before inspecting outcomes, write the question, dataset, variables, preprocessing, baseline, method, metric and expected visualization.
Research question and hypothesis
A research question names a population, measurable variables and a possible relationship.
Responsible biomedical AI
A dataset can underrepresent groups and produce uneven errors.
Sample metadata and batch labels
Where should collection batch be recorded?
Samples and biological features
A data matrix organizes rows as observations and columns as features.
Samples, features and targets
Supervised learning uses known targets to learn a prediction from features.
Scientific communication and defense
A concise research account moves from question to method, result, interpretation, limitation and next experiment.
Scientific interpretation and defense
Which sequence best structures a short defense?
Scientific Python and reproducibility
A notebook executes cells in the order you run them, which may differ from their position on screen.
Sensitivity, specificity and imbalance
Which metric asks what fraction of actual cases were found?
Sequence representation and mutation
DNA is a string over A, C, G and T.
Spike-train representation
A spike train is a list of event times.
Standard deviation units
Which quantity retains the measurement unit?
Subgroup fairness and privacy
Which classroom data choice avoids patient privacy exposure?
Temporal coding and firing rate
Three spikes in one second correspond to what rate?
Toggle switches and oscillators
A toggle switch uses mutually repressing genes to support alternative states; an oscillator uses delayed negative feedback to create repeated changes.
Toggle-switch interactions
A toggle switch mainly uses which interaction?
Train/test contamination
Two recordings from one patient should be split how?
Uncertainty and inference
A sample estimate has uncertainty because a different sample might give a different result.
Variables, types and units
Which name makes a 2.4 mg/mL concentration easiest to audit?