BMS5021 · Introduction to Bioinformatics
The Analysis Workflow & Scientific Reporting
This cross-cutting chapter of Monash University BMS5021 is the assessment method itself: how a bioinformatics analysis is assembled end-to-end and written up. It threads the whole pipeline together - raw data, quality control, alignment, quantification, interpretation - and covers reading your own results critically and structuring the data-analysis report that both Topic 2 and Topic 3 are marked on. Because 70% of the unit is written reports (35% + 35%), the skills here - a reproducible pipeline, justified steps, clear figures - are worth the most marks in the unit.
What this chapter covers
- 01The end-to-end pipeline: raw data -> QC/trim -> align/map -> quantify (count/normalise) -> statistical test -> interpret
- 02Reproducibility: record every tool, version, parameter and step so the analysis can be repeated
- 03Reading your own results critically: sanity-check alignment rates, normalisation and significance before drawing conclusions
- 04Report structure: introduction, methods, results with labelled figures, and justified conclusions
- 05Interpreting mapping statistics: reads aligned 0x / exactly 1x / >1x and the overall alignment rate
- 06Every figure labelled; every analytical choice (filter, normalisation, threshold) justified
- 07The health-data life cycle (ULO2): samples + metadata -> omics -> QC -> feature extraction -> modelling -> insight
- 08Elementary statistics and informative graphical displays of omics data (ULO6)
Order the pipeline and read the alignment stats (assessment-method reasoning)
- +1Order the pipeline. The correct sequence is: (1) quality-control and trim the raw reads, (2) align to the reference, (3) index/sort the alignment (sort then index the BAM), (4) quantify reads per gene (e.g. featureCounts), (5) differential-expression test (edgeR). QC always comes first; you cannot count before you align, and you sort/index before counting.
- +1Compute and judge the alignment rate. The overall alignment rate is the fraction of reads that aligned at least once = 47.23% (exactly once) + 41.77% (more than once) = 89.00% (equivalently 100% - 11.00% that aligned 0 times). An ~89% overall alignment rate is a healthy result for RNA-seq, indicating a good library and the right reference; the ~42% multi-mapping reads reflect repeated/shared sequence and are handled at the counting step.
- +1Name a reporting expectation. A marker rewards, for example, a reproducible methods section that records the tools, versions and key parameters (or any of: clearly labelled figures, and each analytical choice - QC thresholds, normalisation, significance cut-off - explicitly justified rather than left implicit). The point is that HOW you report - transparently and reproducibly - is assessed, not just the final numbers.
Key terms
- Bioinformatics pipeline
- The ordered chain of analysis steps from raw data to interpreted result: QC/trim -> align/map -> sort/index -> quantify -> statistical test -> interpret. Each stage feeds the next, so order matters.
- Reproducibility
- Recording every tool, version, parameter and step so another analyst (or you, later) can repeat the analysis and get the same result - a core expectation of the data-analysis reports.
- Overall alignment rate
- The percentage of reads that mapped to the reference at least once (aligned exactly once + more than once, or 100% minus those aligning zero times); a key sanity-check on library and reference quality.
- Multi-mapping reads
- Reads that align to more than one genomic location because of repeated or shared sequence; they are flagged at alignment and handled carefully at the read-counting step.
- Health-data life cycle (ULO2)
- The flow from biological samples plus metadata, through high-throughput omics, QC, feature extraction and modelling, to a clinical report or insight - the framing that motivates the whole unit.
- Data-analysis report
- The written deliverable for Topics 2 and 3 (35% each): a structured account of the analysis with methods, labelled figures and justified conclusions, on which most of the unit's marks depend.
The Analysis Workflow & Scientific Reporting FAQ
Why does this chapter matter so much for the mark?
Because BMS5021 is assessed 70% by written data-analysis reports (Topic 2 and Topic 3, 35% each) plus a presentation, and those are marked on HOW you assemble and report the analysis, not only on getting a number. A reproducible pipeline, justified choices and clear figures are what separate a High Distinction report from a merely correct one. Confirm the specific rubric for each report on Moodle.
What makes a bioinformatics analysis reproducible?
Recording enough that someone else could repeat it: the tools and their versions, the exact parameters, the reference used, and the order of steps. In a report that means a methods section a reader could follow to reproduce your alignment rate and DE results. Reproducibility is both good science and directly rewarded, so keep a running log of commands and settings as you work rather than reconstructing them at the end.
How do I know if my results are sensible before I write them up?
Sanity-check each stage. Is the overall alignment rate reasonable (an ~89% RNA-seq rate is healthy)? Did filtering and normalisation behave as expected? Are the significant genes credible on the FDR, and does the volcano plot look right? Reading your own results critically - and reporting the checks - is exactly the skill the reports assess. If something looks off, trace it back through the pipeline before drawing conclusions.
Can AI help me structure and check my report?
Yes. Sia can help you plan a clear report structure, pressure-test whether your methods are reproducible, check that each analytical choice is justified, and explain any pipeline step you are unsure about - step by step. It does not write the graded report for you or fabricate results, and Monash University academic-integrity rules apply; because AI use in the reports is governed per task by the educator, follow the acknowledgement rules on Moodle.
Exam move
Since most of the unit's marks live in the two reports, practise the whole workflow as a narrative you can reproduce, not a set of isolated tools. Learn the pipeline order cold (QC/trim -> align -> sort/index -> quantify -> test -> interpret) and keep a running log of every tool, version and parameter as you work so your methods are reproducible. Build the habit of sanity-checking each stage - alignment rate, normalisation, FDR significance - and reporting those checks. When you write, justify every analytical choice and label every figure, because that is what markers reward in a 70%-report unit. Rehearse the Topic 2 presentation too. Ask Sia to quiz you on pipeline order and to critique a draft methods section for reproducibility and justification. Confirm each report's rubric and submission rules on Moodle.
Working through The Analysis Workflow & Scientific Reporting in BMS5021? Sia is AskSia’s AI Biology tutor — ask any BMS5021 The Analysis Workflow & Scientific Reporting 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.
