Bioinformatics & Computational Biology
Chapter: 6.4 Bioinformatics & Computational Biology Level: 1 — Recognition Time limit: 20 minutes Total marks: 30
Section A — Multiple Choice (1 mark each)
Choose the single best answer.
Q1. Which best defines bioinformatics? (1) A. The wet-lab cloning of genes B. The application of computational tools to store, retrieve and analyse biological data C. The microscopic study of cells D. The industrial fermentation of microbes
Q2. Which database is primarily a repository of nucleotide sequences? (1) A. UniProt B. PDB C. GenBank D. BLOSUM
Q3. The Protein Data Bank (PDB) primarily stores: (1) A. Raw sequencing reads B. 3-D atomic coordinates of macromolecular structures C. Phylogenetic trees D. Gene expression counts
Q4. BLAST is used to: (1) A. Predict protein 3-D structure from scratch B. Search a database for sequences similar to a query C. Call variants from a BAM file D. Normalise RNA-seq counts
Q5. A key difference between PAM and BLOSUM matrices is that: (1) A. Higher PAM numbers represent more divergent sequences, while higher BLOSUM numbers represent more similar sequences B. Both increase with divergence C. BLOSUM is for DNA and PAM is for RNA D. They are numerically identical
Q6. Which method builds a phylogenetic tree by repeatedly joining the closest pair of taxa using a distance matrix? (1) A. Maximum likelihood B. Neighbour-joining (UPGMA-type distance method) C. BLAST D. Gene prediction
Q7. AlphaFold is a tool for: (1) A. Multiple sequence alignment only B. Predicting protein 3-D structure from amino acid sequence C. Variant calling D. Building GenBank records
Q8. In a variant-calling pipeline, reads are typically first aligned to a: (1) A. Phylogenetic tree B. Reference genome C. Scoring matrix D. Protein structure
Q9. In RNA-seq analysis, "differential expression" refers to: (1) A. Errors introduced during PCR B. Genes whose transcript levels differ significantly between conditions C. The 3-D fold of a protein D. The number of exons in a gene
Q10. Ab initio gene prediction relies mainly on: (1) A. Comparison to known homologues only B. Intrinsic signals such as start/stop codons, splice sites and codon usage C. Protein crystallography D. BLOSUM scores
Section B — Matching (1 mark each; 6 marks)
Q11–Q16. Match each term in Column X with its correct description in Column Y. Write the letter. (6)
| Column X | Column Y | |
|---|---|---|
| Q11. UniProt | A. Aligns 3+ sequences simultaneously | |
| Q12. Multiple sequence alignment | B. Curated protein sequence & function database | |
| Q13. E-value (BLAST) | C. Visual genome browser track display | |
| Q14. Machine learning | D. Number of hits expected by chance | |
| Q15. FastQC / genome browser | E. Algorithms that learn patterns from data | |
| Q16. Homology | F. Shared ancestry inferred from sequence similarity |
Section C — True / False with Justification (2 marks each; 14 marks)
State True or False (1) and give a one-line justification (1).
Q17. A lower BLAST E-value indicates a more statistically significant match. (2)
Q18. Global alignment aligns the entire length of two sequences, whereas local alignment finds the best matching subregions. (2)
Q19. BLOSUM62 is derived from blocks of conserved sequences and is a widely used default for protein searches. (2)
Q20. UPGMA assumes a constant rate of evolution (a molecular clock) across all lineages. (2)
Q21. In RNA-seq, read counts must always be used raw without any normalisation before comparing samples. (2)
Q22. Supervised machine learning requires labelled training data. (2)
Q23. A VCF file stores predicted protein 3-D structures. (2)
Answer keyMark scheme & solutions
Section A (10 marks)
Q1. B — Bioinformatics = computational storage/retrieval/analysis of biological data. (1)
Q2. C — GenBank is NCBI's nucleotide sequence database; UniProt=protein, PDB=structure. (1)
Q3. B — PDB holds 3-D atomic coordinates from X-ray/NMR/cryo-EM. (1)
Q4. B — BLAST = Basic Local Alignment Search Tool, finds similar sequences in a database. (1)
Q5. A — PAM: higher number = greater evolutionary distance; BLOSUM: higher number = higher % identity blocks (more similar). (1)
Q6. B — Neighbour-joining/UPGMA are distance-based clustering methods joining closest pairs. (1)
Q7. B — AlphaFold predicts 3-D structure from sequence using deep learning. (1)
Q8. B — Reads are aligned to a reference genome before variant calling. (1)
Q9. B — Differential expression compares transcript abundance across conditions. (1)
Q10. B — Ab initio uses intrinsic signals (codons, splice sites, GC content), not homology. (1)
Section B (6 marks)
| Q | Answer | Why |
|---|---|---|
| Q11 | B | UniProt = curated protein database |
| Q12 | A | MSA aligns three or more sequences |
| Q13 | D | E-value = expected chance hits |
| Q14 | E | ML learns patterns from data |
| Q15 | C | Browsers display genome tracks |
| Q16 | F | Homology = shared ancestry |
(1 mark each; all-or-nothing per item.)
Section C (14 marks)
Q17. True (1) — Lower E-value → fewer chance hits → more significant. (1)
Q18. True (1) — Global (e.g. Needleman–Wunsch) aligns full length; local (e.g. Smith–Waterman) finds best subregions. (1)
Q19. True (1) — BLOSUM62 built from conserved BLOCKS at ~62% identity; standard default. (1)
Q20. True (1) — UPGMA assumes equal evolutionary rates (ultrametric/molecular clock). (1)
Q21. False (1) — Counts must be normalised (e.g. library size, TPM/CPM) before valid comparison. (1)
Q22. True (1) — Supervised learning trains on input–label pairs. (1)
Q23. False (1) — VCF stores variant calls (positions/genotypes); structures are in PDB. (1)
[
{"claim":"Section A has 10 one-mark questions","code":"result = (10*1 == 10)"},
{"claim":"Total marks = 10 (A) + 6 (B) + 14 (C) = 30","code":"result = (10 + 6 + 2*7 == 30)"},
{"claim":"Section C: 7 questions at 2 marks = 14","code":"result = (7*2 == 14)"},
{"claim":"Total questions = 23","code":"result = (10 + 6 + 7 == 23)"}
]