Bioinformatics & Computational Biology

Q1. (4 marks)

• Bioinformatics = the application of computational and statistical methods to store, retrieve, analyse and interpret biological data, especially molecular sequence and structure data. (2 marks)
• Any two goals (1 mark each): organise biological data in accessible databases; develop tools/algorithms for data analysis; interpret data to gain biological insight (e.g., predict gene/protein function, evolutionary relationships). (2 marks)

Q2. (3 marks — 1 each)

• (a) GenBank → nucleotide sequences
• (b) UniProt → curated protein sequences and function
• (c) PDB → protein 3D structures

Q3. (4 marks)

• Pairwise alignment: aligns two sequences to find regions of similarity. (1) Use: detecting homology / measuring similarity between two sequences (e.g., BLAST hit). (1)
• Multiple sequence alignment (MSA): aligns three or more sequences simultaneously. (1) Use: identifying conserved motifs/domains, building phylogenetic trees. (1)

Q4. (5 marks)

• BLAST = Basic Local Alignment Search Tool. (2)
• Homology search: finds sequences in a database that are similar to a query, inferring evolutionary relationship / function by similarity. (2)
• Protein query vs protein database program: BLASTP. (1)

Q5. (5 marks)

• BLOSUM (BLOcks SUbstitution Matrix): derived from observed substitutions in conserved blocks of aligned sequences. (1) PAM (Point Accepted Mutation): derived from mutations in closely related sequences and extrapolated over evolutionary distance. (1)
• Higher BLOSUM number = more similar sequences; higher PAM number = more divergent sequences. (1)
• BLOSUM62 vs BLOSUM45: BLOSUM62 is more suitable for closely related sequences (higher number → shorter evolutionary distance). (1) Justification: higher-numbered BLOSUM matrices are built from more identical/conserved blocks. (1)

Q6. (4 marks) Position-by-position:

• M–M identical → +4
• K–R mismatch → −1
• L–L identical → +4
• V–V identical → +4
• A–A identical → +4

Sum = 4 − 1 + 4 + 4 + 4 = 15. (4 marks; 2 for correct per-position scores, 2 for correct total)

Q7. (4 marks)

• Two methods (1 each): Distance-based (UPGMA, Neighbour-Joining); Character-based (Maximum Parsimony, Maximum Likelihood, Bayesian). (2)
• Principle of one (2): e.g., Neighbour-Joining builds a tree by iteratively joining the pair of taxa that minimises total branch length, using a distance matrix. (2)

Q8. (3 marks — 1 each, order matters)

1. Quality control / read trimming (raw FASTQ reads)
2. Alignment/mapping to reference genome (or transcript quantification)
3. Differential expression analysis (counting reads per gene → statistics)

Q9. (4 marks)

• AlphaFold is a deep-learning (AI) system developed by DeepMind that predicts the 3D structure of a protein from its amino acid sequence. (2)
• It addresses the protein folding / structure prediction problem — determining how a linear sequence folds into a functional 3D shape without needing experimental methods like X-ray crystallography. (2)

Q10. (4 marks — any three stages, ~1.3 each)

1. Read alignment/mapping to a reference genome.
2. Post-processing (mark duplicates, base quality recalibration).
3. Variant calling (identify SNPs/indels differing from reference).
4. Variant filtering/annotation (produce final VCF). (4)

[
  {"claim":"Q6 alignment total score = 15 (match +4, mismatch -1)","code":"scores=[4,-1,4,4,4]; total=sum(scores); result = (total==15)"},
  {"claim":"Q6 has 4 identical matches and 1 mismatch","code":"s1=['M','K','L','V','A']; s2=['M','R','L','V','A']; matches=sum(1 for a,b in zip(s1,s2) if a==b); mism=sum(1 for a,b in zip(s1,s2) if a!=b); result = (matches==4 and mism==1)"},
  {"claim":"Recomputed score from sequences equals 15","code":"s1=['M','K','L','V','A']; s2=['M','R','L','V','A']; total=sum(4 if a==b else -1 for a,b in zip(s1,s2)); result = (total==15)"}
]