AI summary by TwinMind
1A. Intro 1: Computational Side of Computational Biology. Statistics; Perl, Mathematica
MIT OpenCourseWare
1h 2m 36s
Course Administration and Structure
- Course numbers: Bio E101, Biophysics 101, HST 508, Genetics 224 - truly interdisciplinary
- Two locations: 5:30pm main campus, noon medical school campus (Cannon room)
- Video backup available for distance education students and absent students
- Grading: 6 problem sets + course project (publication-grade computational biology)
- Collaboration encouraged with source acknowledgment required
- Suzanne Camille: head teaching fellow, most TFs took course previously
- Questionnaires due immediately after class for section assignment
- Schedule accommodates both Harvard and MIT calendars
Course Overview and Topics
- Introductory lectures: Computing and biology foundations
- Central dogma series: 2 lectures each on DNA, RNA, proteins/proteomics
- Reflects information flow: DNA → RNA → proteins
- Enables functional genomics and data mining
- Systems biology focus: 3 lectures on networks (cellular to ecological scale)
- Student presentations: 3 two-hour sessions for team/individual project presentations
Living vs Computational Systems Theme
- Discrete vs continuous data modeling
- Self-assembly and replication as unique living system properties
- Minimal examples approach for illustrating key concepts
- Catalysis and replication using differential equations
- Directed graphs for growth/pedigree analysis
- Single molecule analysis requiring different statistical approaches
Quantitative Biology and Life Definition
- Probability of replication rather than binary alive/not alive
- Complexity from simplicity using environmental components
- Faithfulness of replication as key metric
- Environmental robustness and adaptability measures
- Challenging examples: mules, fire, crystals, viruses, predators requiring complex environments
Complexity Definitions (4 types used in course)
- Computational complexity: Speed/memory tradeoffs, scaling behavior
- Algorithmic complexity/randomness: Minimal bits needed after redundancy removal
- Entropy/information: Bits needed for full compression (Shannon/Boltzmann fusion)
- Physical complexity: Between highly ordered (low entropy) and random (high entropy)
- Complex systems have multiple symmetries and structural scales
- Cannot be represented as simple ordered systems or random coin tosses
Modeling Rationale
- Understanding biological/chemical data through successive approximation
- Designing useful system modifications and testing predictions
- Data sharing - databases contain models, not raw data
- Searching capabilities more powerful with models than raw data
- Data merging/integration requires alignment and redundancy removal
- Error checking - embrace outliers for discovering data/model errors or new findings
- Integration challenges - connecting multiple data types is major current challenge
Error Types and Precision
- Random errors: Average out over repeated experiments
- Systematic errors: Persist across repetitions, require paradigm changes
- Multiple method collection needed to address systematic errors
- SI units: 7 basic units (meters, kilograms, moles, degrees Kelvin, candela, amperes, seconds)
- Precision limits: Physics can achieve 14 significant figures, biology typically 3-4
- Quantum limits: Mole (6×10²³ entities) important for single molecule studies
Minimal Life Examples
- 5-element system: C, H, N, O, P in simple environment (water, ammonium, nucleotide triphosphates, lipids)
- RNA-based life: RNA has catalytic capabilities, can be early replicating unit
- Hypercycle example: Trinucleotides → hexanucleotides with mutual catalysis
- ccg + ccg → ccgccg (catalyzed by cggcgg)
- Creates exponential auto-catalytic cycles
- High replication probability, faithful, but low complexity
Programming Languages and Precision
- Pearl and Mathematica chosen for course (easiest learning curve)
- Precision comparison: Different languages handle transcendental numbers differently
- Mathematica supports arbitrary precision (demonstrated with 100-digit e)
- Other languages show trailing zeros, internal representation limitations
- Pearl: Open source, strong for web applications
- Mathematica: Strong symbolic/numeric math, graphics capabilities
Biological-Computational Analogies
- Instruction sets: Programs (computers) vs Genomes (organisms)
- Bits: 0,1 (computers) vs A,C,G,T (biology)
- Memory hierarchy: Disk/tape (stable) vs RAM (active) // DNA vs RNA
- Environment: Complex (computers) vs can be simple (organisms)
- Input/Output: Analog→Digital→Analog processing in both systems
- Replication: Factories vs cells (femtoliter volumes, ~1 cubic micron)
Virus Examples and Societal Impact
- Computer viruses: Cost 4× entire Human Genome Project budget annually
- AIDS virus: 20 million deaths, worse than Black Plague/1918 influenza
- Replication mechanisms: Copy commands in computer viruses vs polymerase in HIV
- Drug resistance: Highlighted mutations in HIV polymerase causing treatment resistance
- Need for vaccines and public health measures
Data Types and Modeling Approaches
- Sequence modeling: Dynamic programming, ancestry assumptions, faithfulness key
- 3D structure: Motifs, catalysis, complementary surfaces, energy/kinetics
- Functional genomics: Clustering, co-regulation analysis
- Systems biology: Boolean (discrete) vs continuous differential equations vs stochastic models
- Optimization: Linear programming for network optimization, biotech applications
