BMS5021 · Introduction to Bioinformatics
Biological Databases & Annotation
Week 8 of Monash University BMS5021 closes Topic 2 by asking how you find and annotate sequences and, critically, how you EVALUATE the databases you use - the focus of ULO4. It covers the major repositories (NCBI, Ensembl, UniProt and the multi-omic cancer portals), sequence-similarity search with BLAST and what an E-value means, and functional annotation of genes and transcripts. Week 8 is also when the Topic 2 report and presentation are delivered, so this chapter both supplies annotation skills and marks the assessment hand-in.
What this chapter covers
- 01Databases are central to bioinformatics; you must evaluate their merits and limitations for a given problem (ULO4)
- 02Major sequence/annotation databases: NCBI, Ensembl, UniProt (protein); primary vs secondary databases
- 03Multi-omic cancer portals named in the unit: TCGA, ICGC, CPTAC (largest proteomic portal), DepMap, COSMIC, cBioPortal, GDC
- 04BLAST / PSI-BLAST / FASTA are HEURISTIC homology-search tools - fast, approximate similarity search
- 05The E-value: the number of hits of at least this score expected by chance in a database of this size; lower E-value = more significant (not derived as a formula here - confirm specifics on Moodle)
- 06Functional annotation attaches meaning (gene name, function, GO terms) to a matched sequence
- 07Gene Ontology (GO): Cellular Component, Molecular Function, Biological Process categories
- 08GO enrichment / over-representation: find GO terms occurring more than expected in a gene set (Fisher-like test), ranked by FDR / fold-enrichment
Identifying an unknown sequence and evaluating the database (Week 8, ULO4)
- +1Choose the tool. You would use a sequence-similarity (homology) search tool such as BLAST (or FASTA/PSI-BLAST). These are HEURISTIC methods - fast and approximate, using seeds and shortcuts to search huge databases quickly - not the exact dynamic-programming alignment used for two short sequences.
- +1Interpret the E-values. The E-value is the number of hits at least this good expected by chance in a database of this size, so a SMALLER E-value means the match is less likely to be random. E = 1e-40 is far more significant (essentially never expected by chance) than E = 2 (about two such hits expected by chance), so the 1e-40 hit is the trustworthy match. (The unit names the E-value conceptually rather than deriving it as a formula - confirm any threshold your task expects on Moodle.)
- +1Evaluate the database (ULO4). For example: a merit of a curated protein database such as UniProt is high-quality, manually reviewed functional annotation; a limitation is that it may lag behind or omit newly discovered or organism-specific entries. The general point ULO4 rewards is matching the database's coverage, curation level and update frequency to your specific biomedical question rather than defaulting to one database.
Key terms
- BLAST
- Basic Local Alignment Search Tool, a fast HEURISTIC method for searching a database for sequences similar to a query. It trades the guaranteed optimality of dynamic programming for the speed needed to search huge databases.
- E-value
- The Expect value: the number of database hits scoring at least as high as the observed match that would be expected by chance, given the database size. A smaller E-value means a more significant, less-likely-by-chance match.
- NCBI / Ensembl / UniProt
- Major public repositories: NCBI and Ensembl host sequence and genome-annotation data; UniProt is the reference database of protein sequence and functional annotation. Choosing between them is a ULO4 evaluation skill.
- Gene Ontology (GO)
- A controlled vocabulary tagging characterised genes with terms in three categories: Cellular Component (where the protein is), Molecular Function (what it does molecularly) and Biological Process (the pathway/process it belongs to).
- GO enrichment analysis
- Testing whether GO terms occur more often in a gene set than expected in a random equal-size set, using a Fisher-like test; results are ranked by FDR and fold-enrichment to interpret what a set of DE genes has in common.
- Multi-omic cancer portals
- Large integrative databases named in the unit - TCGA, ICGC, CPTAC (the largest proteomic portal), DepMap, COSMIC, cBioPortal and GDC - that combine genomic, transcriptomic and proteomic data for cancer research.
Biological Databases & Annotation FAQ
How is Week 8 assessed?
Two ways. First, database evaluation and annotation is examinable Topic 2 content and maps to ULO4 (evaluating the merits and limitations of databases) and ULO5 (using tools and databases). Second, Week 8 is when the Topic 2 report and in-class presentation are delivered - so it is also the assessment hand-in for the 35% RNA-seq task. Confirm the exact submission and presentation timing on Moodle.
Why is BLAST called heuristic if it works so well?
Because it uses shortcuts - short seed matches and scoring heuristics - to avoid the full, exact dynamic-programming alignment against every database entry, which would be far too slow. Those shortcuts make it fast enough to search enormous databases, at the small risk of missing a true but weak match. It is 'good enough and fast', which is exactly what database search needs.
What does the E-value actually tell me?
It tells you how surprised to be by a hit. The E-value is the number of matches at least this strong you would expect to see by chance in a database of that size, so a very small E-value (say 1e-40) means the match is almost certainly real, while an E-value around 1 or more means a hit that good is expected by chance. The unit treats the E-value conceptually rather than deriving its formula; confirm any threshold your assessment specifies on Moodle.
How do I answer a ULO4 'evaluate the database' question?
Frame it as fit-for-purpose: state the specific biomedical question, then weigh each candidate database on coverage, curation quality and update frequency, and pick the one whose strengths match the question while naming its limitations. For example, a curated protein database gives trustworthy annotation but may omit the newest entries. Sia can help you rehearse this merit-versus-limitation framing for named databases, and it checks your reasoning rather than writing the answer for you.
Exam move
Split your prep between the practical (BLAST and annotation) and the evaluative (ULO4). Practise running a BLAST search on an unknown sequence and reading the hit table - especially the E-value, remembering smaller = more significant - and following a hit through to its functional annotation and GO terms. For the evaluation skill, build a short comparison of NCBI, Ensembl and UniProt (and be aware of the cancer portals TCGA/CPTAC/COSMIC) on coverage, curation and update frequency, and rehearse justifying a database choice as merit-versus-limitation for a stated problem. Because the Topic 2 report and presentation land this week, make sure your annotation and interpretation are report-ready. Ask Sia to quiz you on E-value interpretation and database fit. Confirm all deadlines on Moodle.
Working through Biological Databases & Annotation in BMS5021? Sia is AskSia’s AI Biology tutor — ask any BMS5021 Biological Databases & Annotation question and get a clear, step-by-step explanation grounded in how BMS5021 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.
