bioinformatics final exam condensed focusing on slides he told us to focus on

Heuristic search

A rapid, approximate search strategy that favors speed over guaranteed optimal alignment.

Why BLAST uses heuristics

Allows extremely fast identification of local similarities in huge databases.

Seed-and-extend method

BLAST finds short matching “seeds” then extends them outward to form alignments.

Local alignment

Focuses on matching the most similar regions between sequences, not the entire length.

BLAST purpose

Finds homologous sequences and identifies evolutionary or functional relationships.

E-value meaning

Number of expected random matches; a lower E-value means a more significant hit.

Importance of E-value

Used to distinguish biologically meaningful matches from random noise.

EXPASY definition

A bioinformatics portal linking major protein tools including UniProt, Prosite, and PDB.

UniProt importance

Primary resource for protein sequence, structure, function, variants, and domain information.

Swiss-Prot vs TrEMBL

Swiss-Prot is manually reviewed; TrEMBL is computationally annotated.

Prosite definition

Database of protein domains, motifs, and profiles used for identifying protein families.

Functional domains

Conserved protein regions responsible for specific biochemical functions.

PDB definition

Database of 3D protein structures determined by X-ray, NMR, cryo-EM, or predicted (AlphaFold).

When PDB is used

To examine active sites, domain organization, and ligand interactions.

BLASTN purpose

Compares nucleotide sequences to identify genes, family members, or splice variants.

BLASTX purpose

Translates DNA → protein to detect potential coding regions.

BLASTP purpose

Compares protein sequences to find orthologs, paralogs, and conserved domains.

Importance of annotation

Makes BLAST results interpretable by showing coding regions, domains, and features.

Alternative splicing in BLAST

Seen when family members align strongly but have different exon structures.

Systems biology definition

An interdisciplinary approach that models entire biological systems rather than single components.

Holistic property

System behavior emerges from interactions, not isolated parts.

Interdisciplinary nature

Integrates biology, math, computing, engineering, and statistics.

Emergent properties

New functions or behaviors that arise from interacting components.

Positive feedback

Amplifies system output and can push systems toward new states.

Negative feedback

Stabilizes systems and maintains homeostasis.

Chaos definition

System behavior highly sensitive to initial conditions; long-term outcomes unpredictable.

Attractors

Stable system states toward which dynamic systems tend to move.

Tipping point

Critical threshold where a small input causes a large system transition.

Why modeling is needed

Systems are too complex for intuition alone; require mathematical simulation.

Data integration importance

Combining omics datasets gives a complete picture of system behavior.

Genomics role

Provides the blueprint of potential cellular functions.

Transcriptomics role

Shows which genes are actively expressed.

Proteomics role

Shows functional proteins actually produced.

Metabolomics role

Reveals biochemical activity and pathway flux.

GLP-1 pathway modeling

Predicts metabolic effects, glucose levels, and drug response profiles.

Modeling dosage effects

Helps determine therapeutic windows and toxicity boundaries.

Metagenomics definition

Sequencing DNA directly from the environment to study entire microbial communities.

Metagenomics advantage

Reveals diversity beyond what can be cultured.

Properties of complex systems

Self-regulating, modular, feedback-driven, capable of emergent behavior.

Evolution definition

Change in genetic composition of populations over generations.

Descent with modification

Organisms inherit traits but accumulate changes over time.

Mutation’s role

Source of all new genetic variation in evolution.

Gene flow

Migration of individuals moves alleles between populations.

Genetic drift

Random fluctuations in allele frequency; strongest in small populations.

Natural selection

Differential reproductive success based on heritable traits.

Gradualism

Slow, continuous evolutionary change.

Punctuated equilibrium

Rapid evolutionary bursts followed by long periods of stasis.

Stasis

Species remain relatively unchanged over long time periods.

Allopatric speciation

Formation of new species in geographically isolated populations.

Peripatric speciation

Small peripheral populations evolve quickly due to strong drift.

Sympatric speciation

New species form without geographic isolation.

Phylogenetic tree

Diagram showing evolutionary relationships based on shared ancestry.

Cladogram

Shows branching order based on shared derived traits.

Distance method steps

Align sequences → count differences → cluster closest taxa → build tree.

Parsimony definition

Tree requiring the fewest evolutionary changes is preferred.

Why parsimony works

Assumes simplest explanation is most likely correct.

Informative site

Position where two taxa share a derived state different from the outgroup.

Non-informative site

Position identical across taxa or showing ambiguous change.

Molecular clock concept

Genetic differences accumulate at roughly constant rates over time.

Calibrating a molecular clock

Use fossils or known divergence times to calculate mutation rate.

Molecular clock limitations

Rates are not constant; different genes evolve differently; selection pressures exist.

Metagenomics significance

Reveals unculturable organisms and novel pathways.

Protein conservation reason

Functional constraints limit acceptable mutations.

RNA secondary structure evolution

Includes compensatory mutations and stability-driven changes.

DNA → RNA → protein

The fundamental flow of biological information.

Localization signals

Codes directing proteins to specific cellular compartments.

Coding as layers

Multiple “languages” inside a cell: genetic, regulatory, structural, signaling.

Synthetic biology definition

Engineers biological systems using standardized parts and design principles.

Goal of synthetic biology

Make biology programmable and predictable like engineering.

BioBrick concept

Standardized DNA parts designed for modular assembly.

Base vector

A plasmid backbone enabling construction of new BioBrick-compatible vectors.

BioBrick assembly methods

Four-enzyme standard assembly or 3A assembly.

Verification of assembly

Antibiotic selection, colony PCR, sequencing.

Oncolytic virus definition

Engineered virus that infects, replicates in, and kills tumor cells selectively.

Tumor-selective replication

Achieved via deleted virulence genes or tumor-specific promoters.

Therapeutic transgenes

Boost immune activation or promote tumor destruction.

Safety mechanisms

Deletion of immune-evasion genes, use of suicide switches.

Targeting modifications

Surface protein engineering to bind only tumor cell receptors.

Glycosylation enzyme needed

Glycosyltransferase.

Cell fate pathway

Regulatory program determining cell decisions: division, differentiation, apoptosis.

Structural bioinformatics

Computational analysis of protein structures and interactions.

Orphan receptor definition

Receptor with no known ligand; detected via homology search.

Role of AI in biology

Analyzes massive datasets and learns patterns inaccessible to manual methods.

Artificial intelligence definition

Systems that learn patterns from data to make predictions or decisions.

Machine learning types

Supervised, unsupervised, and reinforcement learning.

Deep learning

AI using multilayer neural networks to learn complex features automatically.

Neural network architecture

Input layer → hidden layers → output layer.

Backpropagation

Process of updating model weights based on prediction errors.

RNN definition

Neural network for sequential data with memory of previous inputs.

LSTM definition

RNN variant designed to learn long-range dependencies.

AI in bioinformatics

Predicts splicing, motifs, gene expression, classification, and structural patterns.

Classification model

Assigns inputs to predefined categories (e.g., SignalP).

SignalP example

Classifies proteins as containing or lacking signal peptides.

PCA purpose

Reduces dimensionality and reveals major patterns in high-dimensional data.

Clustering definition

Groups data based on similarity without labels.

Linear regression

Predicts continuous numerical outcomes from input variables.

Logistic regression

Predicts binary outcomes such as disease vs healthy.

Survival analysis

Models time until an event like relapse or death occurs.

Supervised vs unsupervised

Supervised uses labeled data; unsupervised identifies structure without labels.

AlphaFold significance

AI system predicting highly accurate protein 3D structures.