Computational Biology Scientific¶

What you'll learn¶

• How computational biology relates to bioinformatics
• Sequence alignment via dynamic programming (tested)
• Structural biology concepts
• Genomic-scale data processing
• The scientific tooling

Computational biology applies computation and modeling to biological questions — overlapping with bioinformatics but leaning more toward algorithms, structure, and simulation. The alignment scoring example here is run-verified.

Biology as an algorithms problem¶

Sequences, structures, and populations become string, graph, and statistics problems.

Much of computational biology reduces to well-known CS problems:

• Sequence alignment → dynamic programming (edit distance).
• Phylogenetic trees → graph/tree algorithms.
• Protein folding → optimization and simulation.
• Genome assembly → graph problems (de Bruijn graphs).

This is why strong CS fundamentals transfer directly. Let's see the most central one.

Sequence alignment scoring (tested)¶

Sequence alignment scoring in Computational Biology — what it is and when to use it.

Comparing two DNA/protein sequences means finding their best alignment — and the core is a dynamic programming score, essentially edit distance. Here's the minimum edit distance (Levenshtein) between two sequences, pure Python:

def edit_distance(a: str, b: str) -> int:
    """Minimum single-character edits (insert/delete/substitute) to turn a into b."""
    m, n = len(a), len(b)
    dp = [[0] * (n + 1) for _ in range(m + 1)]
    for i in range(m + 1):
        dp[i][0] = i                      # delete all of a's first i chars
    for j in range(n + 1):
        dp[0][j] = j                      # insert all of b's first j chars
    for i in range(1, m + 1):
        for j in range(1, n + 1):
            cost = 0 if a[i-1] == b[j-1] else 1
            dp[i][j] = min(
                dp[i-1][j] + 1,           # deletion
                dp[i][j-1] + 1,           # insertion
                dp[i-1][j-1] + cost,      # match or substitution
            )
    return dp[m][n]

print(edit_distance("GATTACA", "GATGCA"))
print(edit_distance("AAAA", "AAAA"))

Output:

2
0

GATTACA → GATGCA needs 2 edits; identical sequences need 0. This dynamic-programming table is the foundation of Needleman-Wunsch (global alignment) and Smith-Waterman (local alignment) — the algorithms that power sequence comparison across all of biology. Real aligners add gap penalties and substitution scoring matrices (like BLOSUM for proteins), but the DP core is exactly this. See Dynamic Programming.

Structural biology¶

Work with 3D molecular structures (proteins, DNA) computationally.

Beyond sequences, computational biology studies 3D structure:

• Protein structure — proteins fold into shapes that determine function. Predicting the fold from the sequence was a grand challenge, recently revolutionized by AlphaFold (deep learning).
• Molecular dynamics — simulating how molecules move and interact over time (continuous simulation — see Computational Physics); tools like OpenMM.
• Structure analysis — measuring distances, angles, and binding sites in 3D coordinates (BioPython's Bio.PDB).

Genomic-scale data¶

Handle datasets too big for memory with streaming and specialized formats.

Modern genomics processes enormous datasets — a single sequencing run produces billions of bases:

• Formats — FASTQ (reads + quality), BAM/SAM (alignments), VCF (variants).
• Pipelines — raw reads → quality control → alignment → variant calling → annotation, orchestrated as a data pipeline (Snakemake, Nextflow).
• Scale — needs efficient tools (pysam), often sparse structures (Sparse Tensors), and sometimes distributed processing.

The ecosystem¶

Biopython, scikit-bio, and friends — the standard toolkit.

Practice exercises¶

1. Modify edit_distance to also return the aligned strings (traceback through the DP table).
2. Add a gap penalty different from the substitution cost and observe how alignments change.
3. Use edit distance to find the closest match of a short sequence within a longer one.
4. Explain why sequence alignment is a dynamic-programming problem (optimal substructure).
5. Research how AlphaFold changed protein structure prediction, and what problem it solved.