Molecular Biology and Bioinformatics Principles

Posted on Aug 16, 2025 in Computer Engineering

I. Molecular Biology Fundamentals

Genes and Genomes

• Every cell contains a complete set of genetic instructions—the genome—encoded in DNA and organized into genes packaged on chromosomes. A gene is a specific DNA sequence that encodes a functional product (usually a protein or an RNA). Genetic variation (mutations) underlies phenotypic differences, while environmental factors also contribute to traits.

DNA vs. RNA

II. Central Dogma of Molecular Biology

Flow of Information:

1. Replication: DNA → DNA

2. Transcription: DNA → RNA

3. Translation: RNA → Protein

Exceptions (e.g., reverse transcription in retroviruses) exist, but generally proteins do not encode nucleic acids.

III. Transcription (DNA → RNA)

• Enzyme: RNA polymerase reads the template strand (3′→5′) and synthesizes RNA (5′→3′).

• Coding vs. Template Strands:

  • Coding strand has same sequence as RNA (T→U).

  • Template strand is complementary and used for synthesis.

• Eukaryotic Processing: 5′ cap, 3′ poly-A tail, and splicing out introns yield mature mRNA.

• Prokaryotes lack a nucleus—transcription and translation are coupled, and mRNAs are used directly without capping or splicing.

IV. Translation (RNA → Protein)

• Ribosome ‘reads’ mRNA codons (three-base units) in the 5′→3′ direction.

• Start codon: AUG (Methionine) establishes reading frame.

• Stop codons: UAA, UAG, UGA terminate synthesis.

• tRNAs with complementary anticodons bring specific amino acids to the ribosome’s A (aminoacyl), P (peptidyl), and E (exit) sites.

• Open Reading Frame (ORF): Region from start to stop codon; six possible frames (3 per strand). Tools like NCBI ORFfinder identify ORFs in a sequence.

V. Genetic Mutations and Variations

Point Mutations (Substitutions):

• Silent (Synonymous): Codon change but same amino acid (e.g., GAA→GAG both code Glu)—no protein change.

• Missense (Nonsynonymous): Different amino acid (e.g., GAG (Glu)→GTG (Val))—effect depends on similarity.

• Nonsense: Codon → stop codon (e.g., TGC→TGA)—truncated protein, often loss of function.

Insertions/Deletions (Indels):

• Frameshift: Indel not a multiple of 3 shifts reading frame downstream, scrambles protein, often leading to an early stop—usually deleterious.

• In-frame: Indel multiple of 3 adds/removes whole amino acids—impact varies by location and size.

VI. Bioinformatics Databases & Tools

Primary vs. Derived Databases:

• Primary (Archive): Raw submissions (GenBank, EMBL, DDBJ, SRA, PDB)—may be redundant or uncurated.

• Derived (Curated): Processed, non-redundant (RefSeq, UniProt/Swiss-Prot, CDD)—standardized, well-annotated.

GenBank vs. RefSeq:

• GenBank: Submission-based, multiple entries per gene, accession.version (e.g., U12345.1).

• RefSeq: Curated single records per biomolecule with stable accessions (e.g., NM_, NP_, NC_).

RefSeq Accession Prefixes:

Key NCBI Tools:

• Entrez: Integrated search across sequence, literature (PubMed), structures.

• ORFfinder: Identifies ORFs in DNA sequences.

• BLAST: Heuristic local alignment search (BLASTN, BLASTP, BLASTX, TBLASTN/TBLASTX) with E-values indicating significance.

• UCSC Genome Browser: Visualize genomic context and annotations.

VII. Sequence Alignment

7.1 Pairwise Alignment

• Dot Plot: Visual match matrix; diagonal lines reveal similarity; adjustable stringency.

• Dynamic Programming:

  • Needleman–Wunsch: Global end-to-end alignment.

  • Smith–Waterman: Local high-scoring segment alignment.

• Scoring Schemes:

  • Matches: Positive scores.

  • Mismatches: Negative scores.

  • Gaps: Opening and extension penalties (e.g., –2, –1).

• Percent Identity: (Identical matches / aligned length) × 100%.

7.2 Substitution Matrices (Proteins)

• PAM Matrices: Based on accepted mutations (PAM1 for ~1% divergence; PAM250 for distant comparisons).

• BLOSUM Matrices: Empirical; e.g., BLOSUM62 for ~62% identity clustering. Choose matrix based on expected divergence.

7.3 BLAST

• Method: Heuristic local alignment via “word” matches extended into High-Scoring Segment Pairs (HSPs).

• E-value: Expected number of random matches; lower values indicate more significance.

• Flavors:

  • BLASTN: Nucleotide vs. nucleotide.

  • BLASTP: Protein vs. protein.

  • BLASTX/TBLASTN/TBLASTX: Translated searches.

VIII. Multiple Sequence Alignment (MSA)

Definition & Purpose

• Aligns three or more homologous sequences simultaneously so each column represents evolutionarily equivalent positions.

• Key applications include identifying conserved motifs/domains, inferring functional sites, preparing input for phylogenetic analysis, and structure prediction.

Challenges vs. Pairwise Alignment

• Exponential search space: aligning n sequences optimally is NP-hard.

• Complexity increases steeply with sequence number and length.

1. Progressive Alignment

• Workflow:

  1. Compute all pairwise distances (e.g., percent identity or substitution-matrix scores).

  2. Construct a guide tree (e.g., UPGMA or Neighbor-Joining) reflecting sequence relationships.

  3. Align the most similar pair of sequences into a profile.

  4. Iteratively align remaining sequences or profiles following the guide tree order.

• Tools: ClustalW, Clustal Omega.

• Advantages: Fast, scalable to large numbers of sequences.

• Limitations: Errors in early pairwise alignments propagate (“once a gap, always a gap”).

2. Iterative Refinement & Consistency Methods

• Iterative Refinement:

  • Repeatedly partition or realign subsets of sequences or the entire alignment to improve score (sum-of-pairs or likelihood).

  • Tools: MUSCLE, MAFFT, PRANK.

• Consistency-Based Alignment:

  • Incorporate information from multiple pairwise alignments to enforce consistency across the MSA.

  • Tools: T-Coffee, ProbCons (uses hidden Markov models for posterior probabilities).

3. Template/Structure-Guided Alignment

• Use known 3D structures to align sequences, preserving structural equivalences.

• Tools: Expresso (3D-Coffee), PROMALS3D.

4. Scoring & Gap Penalties

• Sum-of-Pairs Score: Sum of scores for all residue pairs in each column (using substitution matrices for proteins).

• Profile-Profile Alignment: Aligns profiles (weighted residue frequencies) rather than raw sequences for later steps.

• Gap Penalties:

  • Affine model: penalty = gap_open + (length × gap_extend).

  • Adjust based on expected indel frequencies and structural regions.

5. Output Formats & Visualization

• Formats: CLUSTAL (.aln), FASTA (with gaps), Stockholm (.sto).

• Viewers: Jalview, MSAViewer, UGENE.

6. Quality Assessment & Post-Processing

• Metrics: Column Score (fraction of pairwise matches), GUIDANCE scores, TCS (Transitive Consistency Score).

• Filtering: trimAl, Gblocks—remove poorly aligned or divergent regions before downstream analyses.

Practical Tips

• Choose algorithms based on dataset size, sequence similarity, and computational resources.

• Inspect alignments manually; realign problematic regions.

• Experiment with different substitution matrices (e.g., BLOSUM62 vs. BLOSUM45) and gap penalties.

• Always trim unreliable columns prior to phylogenetic inference or motif discovery.

IX. Phylogenetic Tree Reconstruction

Concepts & Terminology

• Leaf (Tip): Observed sequences.

• Internal Node: Hypothetical ancestor.

• Branch Length: Represents evolutionary change or time.

• Topology: Branching order (unrooted vs. rooted).

• Clade (Monophyletic Group): Ancestor plus all its descendants.

1. Input: MSA of Homologous Sequences

• Quality of MSA directly impacts tree accuracy. Trim ambiguous regions.

2. Models of Sequence Evolution

• Nucleotide Models: Jukes-Cantor, Kimura 2-parameter, General Time Reversible (GTR).

• Protein Models: JTT, WAG, LG—incorporate amino acid substitution frequencies.

3. Tree-Building Methods

A. Distance-Based Methods

• UPGMA (Unweighted Pair Group Method with Arithmetic Mean):

  • Assumes a molecular clock (constant rate); yields an ultrametric (rooted) tree.

  • Simple but unreliable if rates vary.

• Neighbor-Joining (NJ):

  • No clock assumption; fast O(n³); produces an unrooted tree.

  • Widely used for exploratory analyses.

B. Character-Based Methods

• Maximum Parsimony (MP):

  • Seeks the tree minimizing total character changes.

  • No explicit evolutionary model; vulnerable to long-branch attraction.

• Maximum Likelihood (ML):

  • Finds the tree maximizing likelihood given a model of substitution.

  • Computationally intensive; tools: PhyML, RAxML, IQ-TREE.

• Bayesian Inference:

  • Estimates the posterior distribution of trees under a model via MCMC.

  • Outputs clade posterior probabilities and can incorporate relaxed molecular clocks (BEAST, MrBayes).

4. Branch Support & Validation

• Bootstrap Analysis:

  • Resample alignment columns with replacement to create replicates.

  • Reconstruct trees; the percentage of replicates supporting each clade equals the bootstrap value.

  • Values >70% are considered strong support.

• Posterior Probabilities: From Bayesian inference; values >0.95 indicate strong support.

5. Tree Formats & Visualization

• Newick Format: (A:0.1,(B:0.2,C:0.2):0.3);

• Nexus Format: Includes metadata, alignment blocks, and tree blocks.

• Visualization: FigTree, iTOL (Interactive Tree Of Life), Dendroscope, ETE Toolkit.

6. Rooting & Molecular Clocks

• Rooting:

  • Outgroup Rooting: Include a known distant relative.

  • Midpoint Rooting: Place the root at the midpoint of the longest path.

• Molecular Clock Models:

  • Strict Clock: Assumes a constant rate; calibrate with fossil or sampling dates.

  • Relaxed Clock: Allows rate variation among branches.

  • Tools: BEAST for dating analyses.

7. Interpretation & Best Practices

• Topology vs. Branch Lengths: Topology shows relationships; lengths indicate the amount of change.

• Monophyly, Paraphyly, Polyphyly: Understand clade definitions.

• Pitfalls:

  • Poor alignment regions lead to misleading branches.

  • Model misspecification leads to incorrect likelihoods or posterior probabilities.

  • Long-branch attraction especially affects parsimony.

• Recommendations:

  • Use multiple tree-building approaches to compare topologies.

  • Employ adequate substitution models and partitioning (e.g., codon positions).

  • Report support values; collapse poorly supported nodes.

X. Practical Applications and Hands-On Examples

X.1. Transcription & Translation Basics

• MCQs Key Answers: 1) OH group; 2) b, d; 3) UAA stop codon.

• Translation: Identify reading frame, transcribe DNA → mRNA (T → U), translate by AUG start.

  • Example: GAGCCAUGCAUUAUCUAGAUAGUAGGCUCUGAGAAUUUAUCUC → Met-His-Tyr-Leu-Asp-Ser-Arg-Leu.

• Promoter Location: Upstream of the transcription start site.

• Mutations:

  • Upstream insertion (5′ UTR) → no protein change.

  • In-frame substitution (synonymous: AAU ⇄ AAC both Asn) → silent mutation.

  • Frameshift insertion in CDS → premature stop, truncated protein.

  • Stop-codon suppression by mutant tRNA → elongated protein.

X.2. Point Mutations & Splicing

• Deletion Effects:

  • 4-nt deletion → removes ≥1 aa + frameshift → likely truncated, nonfunctional.

  • 3-nt deletion → removes 1 aa; possible amino acid substitution if crossing codons.

• Insertion of 1 nt near start/end → frameshift → high probability of harm.

• Substitution Types:

  1. Silent (synonymous)

  2. Conservative missense (similar property amino acid)

  3. Non-conservative missense (different property amino acid)

  4. Nonsense → early stop.

• Splice-site mutation: Single intronic base change abolishes original acceptor (AG), uses downstream AG → skips exon → mRNA 173 nt shorter.

X.3. Beta-Globin Cluster & Allele Variation

• Codon Translation:

  • atg gtg cac ctg act cct gag gag aag → MVHLTP E E K.

  • Variant at position 7: E → V if codon GAG → GTG.

• Cluster Gene Order (5′→3′): ε – γ-G – γ-A – δ – β.

• Strand Orientation: Reverse-strand genes have complementary designations (A ↔ T).

• Allele Observations: Non-synonymous mutation changes residue; lack of reported allele frequency signals potential data error.

X.4. Gene Structure & Thalassemia

• Gene Anatomy:

  • 5′ UTR – exon 1 – intron – exon 2 – intron – exon 3 – 3′ UTR – poly-A signal/tail.

  • Transcription start site at 5′ end; translation begins at the first AUG in the exon.

• Beta-Thalassemia Mutations:

  1. Nonsense (GAG → TAG in exon) → β⁰, no beta-globin.

  2. Single-nt insertion → frameshift → β⁰.

  3. 4-nt deletion → frameshift → β⁰.

  4. Splice-site (IVS I) G → A → aberrant splicing → β⁺, reduced beta-globin.

X.5. Synonymous Substitutions & Conservative Changes

• Synonymous (Silent) Substitutions: Occur at the third codon position; protein sequence unchanged.

• Transition vs. Transversion:

  • Transition: purine ↔ purine (A ↔ G), pyrimidine ↔ pyrimidine (C ↔ T).

  • Transversion: purine ↔ pyrimidine.

  • Observation: Transitions are more frequent than transversions.

• Conservative Amino Acid Changes:

  • Basic: Arg ↔ Lys; nonpolar: Ile, Val, Leu, Met.

• Phylogenetic Signal: Gaps indicate clade distinctions (Afrotheria vs. Eutheria vs. Marsupials).

X.6. HIV/SIV Phylogenetics

• Data: 9,176 bp HIV-1 vs. SIV sequences; 9 protein-coding genes.

• Tree: Strong bootstrap support shows HIV-1 clusters with SIVcpz, HIV-2 with non-chimpanzee SIVs.

• Genome: ~9,176 bp, 9 coding genes. Strong bootstrap support at key nodes.

• Relationships: HIV-1 clusters with SIVcpz; HIV-2 with non-chimpanzee SIVs.

• Bootstrap Analysis: Percentage of replicates supporting each clade (≥70% indicates strong support).

X.7. Mitochondrial & MYH16 Pseudogene Analysis

• Sequences: Human, Neanderthal, Denisovan, Pan, Gorilla, etc.

• Newick Example:

  ((((Pan_paniscus:0.000001,Gorilla_gorilla:0.000001):0.06,(Homo_sapiens:0.000001,Pongo_pygmaeus:0.000001):0.12):0.12,Pan_troglodytes:0.000001);

• Interpretation: Closest relatives are indicated by shortest branch lengths; divergence is indicated by branch length values.

• Reference Accessions:

  • Human: NC_012920.1; Neanderthal: NC_011137.1; Denisovan: NC_013993.1.

• Percent Identity: Human–Neanderthal ~98.72%; Human–Denisovan ~97.61%.

• Residue Change: Isoleucine (I) → Valine (V) conservative mutation at specified codons.

• Pseudogene 2-base Deletion: BK001410 MYH16 frameshift; truncated pseudogene.

• Phylogenetic Tree (Newick): Indicates evolutionary distances; shortest branches represent closest relatives.

XI. Essential Formulas (with Descriptions)

1. Percent Identity
   
   Percent Identity = (Number of identical aligned positions / Aligned length) × 100%

   • Identical aligned positions: count of positions where residues match exactly.

   • Aligned length: total non-gap columns compared.

2. Percent Similarity
   
   Percent Similarity = (Number of similar (conservative) residue pairs / Aligned length) × 100%

   • Similar pairs: substitutions between biochemically similar residues (e.g., Ile ↔ Val).

3. Transition/Transversion Ratio (κ)
   
   κ = #Transitions (A ↔ G, C ↔ T) / #Transversions (purine ↔ pyrimidine)

4. Affine Gap Penalty
   
   GapPenalty = GapOpen + (L – 1) × GapExtend

   • GapOpen: penalty for introducing a new gap.

   • GapExtend: penalty for each additional position in the same gap.

   • L: length of the gap in residues.

5. Alignment Score (S)
   
   S = Σₖ s(aₖ,bₖ) + Σ_g [GapOpen_g + (L_g – 1) × GapExtend_g]

   • s(aₖ,bₖ): substitution score for aligned residues.

   • G: number of gaps.

   • L_g: length of the g-th gap.

6. Needleman–Wunsch Recurrence (Global Alignment)
   
   F(i,j) = max{ F(i-1,j-1) + s(xᵢ,yⱼ), F(i-1,j) – d, F(i,j-1) – d }

   • F(i,j): best score for prefixes x₁…xᵢ and y₁…yⱼ.

   • s(xᵢ,yⱼ): match/mismatch score.

   • d: gap penalty.

7. Smith–Waterman Recurrence (Local Alignment)
   
   H(i,j) = max{ 0, H(i-1,j-1)+s(xᵢ,yⱼ), H(i-1,j) – d, H(i,j-1) – d }

   • H(i,j): best local score ending at positions.

   • Zero bound ensures restarting at zero when negative.

8. p-distance
   
   p = Number of differing sites / Total sites compared

9. Jukes–Cantor Model Distance
   
   d = -¾ × ln(1 – 4/3 × p)

   • p: observed proportion of differences.

10. Kimura 2-Parameter Distance
    
    d = -½ ln(1 – 2P – Q) – ¼ ln(1 – 2Q)

• P: proportion of transitions.

• Q: proportion of transversions.

11. Bootstrap Support (%)
    
    %Bootstrap = (#Replicates supporting clade / Total replicates) × 100%

12. BLAST E-value
    
    E = K × m × n × e^(−λS)

• m, n: query and database lengths.

• S: raw score.

• λ, K: statistical constants.

13. ORF Count
    6 (three reading frames per strand × two strands)

14. Total Codons
    4³ = 64 possible codons