Writing your first manuscript is not primarily a writing problem. It’s a structure problem. Most people struggle because they try to write linearly when manuscripts are not read linearly and were not meant to be written that way. This tutorial teaches you the structure and the order that actually works.
The Anatomy of a Scientific Manuscript: IMRaD
Scientific papers follow a rigid structure for good reason. It’s called IMRaD: Introduction, Methods, Results, Discussion. Each section accomplishes something specific for the reader.
Introduction (the context). Why does this problem matter? What’s been done before? What’s the gap in knowledge? This section funnels from broad context down to your specific research question.
Methods (the roadmap). What did you do, exactly, so someone could reproduce it? Reagents, concentrations, software versions, statistical tests, everything.
Results (the core). What did you find? This is the paper’s contribution. No interpretation here, just the facts and figures.
Discussion (the meaning). What do these results mean in the context of existing literature? How do they change what we know? What are the limitations? This is where you interpret and speculate.
The key insight: Results is the core of the paper. Everything else exists to contextualize the Results. Readers who skim will read the Abstract and Results; they might skip Discussion. Your Results must stand alone and be comprehensible.
Because Results is the heart, you should write it first.
Writing Order (This Is Counterintuitive But Works)
The conventional advice is to write Introduction first, then Methods, then Results, then Discussion. This is bad advice. You’ll stare at a blank Introduction, not knowing what context to provide until you’ve written Results. You’ll rewrite the Introduction multiple times as Results changes. You’ll procrastinate.
Write in this order instead: Methods, Results, Introduction, Discussion, Abstract.
Why This Order Works
Methods first. You already know what you did. Writing this forces you to organize your experimental design clearly and identify any gaps (did you actually record the statistical test you’re claiming?). Methods is also less intimidating than Introduction, so starting here builds momentum.
Results second. Now you write the core of the paper. Your figures and data are in front of you. You describe what each figure shows and what the numbers mean.
Introduction third. Once you’ve written Results, you know exactly what context your findings need. You can write a targeted Introduction that funnels down to your specific contribution, rather than a rambling literature review that doesn’t connect to your work.
Discussion fourth. You’ve written Results and Introduction, so you know what you found and what’s been done before. Discussing the meaning is now straightforward.
Abstract last. The Abstract summarizes a complete manuscript. Write it after everything else exists.
This order eliminates blank-page paralysis and prevents the circular rewriting that kills momentum.
Methods Section: Be Specific Enough to Reproduce
The Methods section is your chance to be precise. Every detail you omit, a reader trying to reproduce your work will guess wrong.
What to Include
Reagents and materials: List every kit, antibody, and chemical with the manufacturer and catalog number. Include concentrations. Example:
“Cells were stained with anti-CD4 (BD Biosciences, clone RPA-T4, catalog 557707) at 1:100 dilution in PBS containing 2% fetal bovine serum.”
Software and versions: Name every software tool with version number. Example:
“RNA-seq reads were aligned to the human reference genome (GRCh38) using STAR v2.7.9a with default parameters.”
Statistical tests: Name the exact test, the model, and any assumptions tested. Example:
“Differences between groups were analyzed using a two-sided Student’s t-test with equal variances assumed (Levene’s test, p > 0.05). Statistical analyses were performed in R v4.2.1 using the base stats package.”
Do not omit details. If you normalized data to a control, say so. If you excluded outliers, say how (interquartile range method, z-score threshold, etc.). If you performed multiple comparisons, describe the correction (Bonferroni, false discovery rate, etc.).
Common Failure Mode
Methods sections that are too vague are useless. “Cells were stained with antibodies and analyzed by flow cytometry” tells nobody what you did. “Cells were stained with anti-CD4-FITC (BD, clone RPA-T4, 1:100) and anti-CD8-PE (Beckman, clone B9.11, 1:50) in PBS with 2% serum for 20 minutes on ice, then analyzed on a BD FACSCalibur using FlowJo v10” is reproducible.
Results Section: Lead With the Finding, Not the Method
Here’s a fundamental mistake most first-time writers make. They write:
“We performed treatment X and measured tumor volume. The treated tumors measured 150 ± 30 mm^3, while control tumors measured 280 ± 40 mm^3.”
This makes the reader work. They have to parse what the numbers mean. Better:
“Treatment X reduced tumor volume by 46% compared to control (treated, 150 ± 30 mm^3; control, 280 ± 40 mm^3; p < 0.01).”
Lead with the finding. The method is a supporting detail.
Structure of Results
Each paragraph should describe one figure or related set of figures:
- State the finding in the topic sentence
- Describe what the figure shows (axes, populations, groups)
- Provide the quantitative result with statistics
- Add a one-sentence interpretation of what this finding means
Example paragraph:
“CD8+ T cell frequency increased sharply after antigen exposure. Flow cytometry of splenocytes revealed CD8+ cells comprised 12 ± 3% of the lymphocyte gate at baseline and 38 ± 7% by day 7 post-immunization (p < 0.001, Figure 1B). This expansion occurred primarily in the first week, with frequencies plateauing by day 14 (28 ± 5%, not significantly different from day 7). The expansion was antigen-specific; no increase was observed in control animals receiving PBS injection.”
Figures Must Stand Alone
Every figure should be comprehensible without reading the text. The figure legend should explain:
- What the figure shows (the experimental design or measurement)
- The axis labels and their units
- Sample sizes and error bars
- Statistical significance if relevant
Example legend:
“Figure 1B. CD8+ T cell frequency following immunization. Splenocytes were isolated from C57BL/6 mice at indicated timepoints post-immunization with OT-I peptide (n=5 mice per timepoint) or control PBS (n=3 mice). Frequencies of CD8+ cells among all lymphocytes were determined by flow cytometry. Error bars show mean ± SEM. ***, p < 0.001 (two-way ANOVA with Bonferroni post-test). ns, not significant.”
Introduction Section: The Funnel Structure
The Introduction should funnel from broad to narrow, ending with your specific research question.
The Funnel
Paragraph 1 (Broad). Start with the disease, disorder, or biological problem at hand. Example:
“Tuberculosis remains the leading infectious cause of death globally, claiming over 1.2 million lives annually despite the availability of effective antibiotics.”
This is your context. The reader should immediately understand why your work matters.
Paragraph 2 (Narrower). What is known about this problem? Cite key studies that have advanced understanding. Don’t cite everything; choose the most important 3-5 papers.
“Mycobacterium tuberculosis persists in macrophages through multiple mechanisms, including lipid-laden granulomas that sequester bacteria from immune surveillance and first-line antibiotics.”
Paragraph 3 (Narrower still). What is not known? What’s the gap?
“However, the role of lipid-specific T cell responses in controlling latent infection remains unclear. Previous studies have focused on protein antigens; lipid-reactive T cells are only recently recognized as important contributors to TB immunity.”
Paragraph 4 (Your question). What are you asking?
“Here we investigate whether CD1b-restricted lipid-reactive T cells expand following Mycobacterium tuberculosis infection and whether they control bacterial persistence in an in vitro model of granuloma.”
Three to five paragraphs is typical. Do not write a literature review. Write a targeted funnel that leads to your specific question.
Common Mistakes
Introductions that wander. The reader finishes your Introduction and doesn’t understand what question you’re answering. Your Introduction should end with “here we show” or equivalent, and it should be clear.
Introductions that are too long. Your Introduction should be 1.5 to 2 pages, not 5 pages. This is not a review article.
Introductions that over-cite obscure papers. Cite papers that directly advance your narrative. One or two sentences and a citation are enough for each idea.
Discussion Section: Interpret, Don’t Summarize
The most common mistake in Discussion is summarizing Results instead of interpreting them.
Bad: “We found that treatment X reduced tumor volume. This shows that treatment X is effective.”
This adds nothing. The reader already read Results.
Good: “Our findings demonstrate that treatment X reduces tumor burden through a macrophage-independent mechanism. Unlike existing therapies that rely on TLR signaling, our data show sustained efficacy in Myd88-deficient mice, suggesting a distinct pathway. These findings align with recent work showing X directly activates tumoricidal T cell responses (Smith et al., 2024), but extend it by demonstrating efficacy in vivo.”
Interpretation places your work in context, challenges or affirms existing models, and points forward.
The Discussion Structure
Leading paragraph (your key finding and what it means in context). Not a summary of Results, but an interpretation. What does this mean? Why should anyone care?
“Our finding that X directly targets dormant tumor stem cells is significant because the field has largely assumed these cells are inherently resistant to small-molecule drugs. By coupling kinetic modeling with single-cell RNA-seq, we show that X achieves penetration at sufficient concentrations, suggesting that prior failures in clinical trials reflected dose or formulation issues, not a fundamental barrier to targeting.”
How your findings extend or challenge existing work. Cite relevant papers. Do your results agree or disagree? Why?
“These findings extend the model proposed by Johnson et al. (2022), who predicted X’s mechanism based on kinetic simulations, and for the first time provide experimental validation. However, we observed activation of pathway Z in 15% of cells, which Johnson’s model did not predict. This suggests…”
Limitations (be honest). What are the weaknesses of your study? What can’t you conclude?
“A limitation of our work is that we studied X only in mice. The penetration of X in human tumors, which are often larger and less vascularized, is unknown. Additionally, our single-cell analysis used a limited panel of markers and may have missed subpopulations.”
Future directions. What’s the next logical step?
“Future studies should assess X efficacy in patient-derived xenografts and determine whether combination with standard chemotherapy improves outcomes. The unexpected pathway Z activation warrants investigation in human cells.”
Closing (do not end with “further research is needed”). Bring the discussion home. What’s the takeaway?
“Our results indicate that dormant tumor stem cells are pharmacologically accessible, opening new therapeutic avenues in patients with otherwise treatment-resistant disease.”
A strong closing provides closure and direction, not a cop-out.
Abstract Section: Write Last
The Abstract is 150-250 words summarizing the entire paper. Write it after everything else is done.
Structured Abstract Format
Most journals use a structured format:
Background (1-2 sentences). The problem. What’s the clinical or biological need?
“Tuberculosis remains difficult to treat because Mycobacterium tuberculosis persists in macrophage granulomas. Current drugs have limited penetration into these sequestered sites.”
Methods (1-2 sentences). What you did. Keep it brief.
“We developed an in vitro granuloma model and assessed penetration and efficacy of a novel lipophilic antibiotic, TBX-1, combined with standard therapy.”
Results (2-3 sentences with key numbers). Your main findings with quantitative data.
“TBX-1 achieved intragranular concentrations 3-fold higher than standard isoniazid and reduced viable bacteria by 78% ± 12% at 4 weeks, compared to 34% ± 8% with standard therapy alone (p < 0.01).”
Conclusions (1-2 sentences). What this means for the field or patients.
“These findings suggest TBX-1 warrants clinical evaluation as an adjunctive therapy for drug-resistant TB. Our model provides a platform for rapid assessment of new granuloma-penetrating agents.”
Keep the Abstract under 250 words and ensure it can stand alone without the paper.
Practical Process Tips
Getting Unstuck
If you’re staring at a blank page, write badly. Your first draft does not need to be good. Write sentences that are clunky, paragraphs that wander, explanations that are unclear. Get words on the page. You’ll revise later. Waiting for perfect sentences to appear is the enemy of progress.
Set a timer for 30 minutes and write continuously, even if you’re not happy with what’s appearing. Stop the timer, read what you wrote, then revise. Revision is easier than creating from scratch.
Working With Your Supervisor or Co-Authors
Do not write the entire manuscript in silence and present it finished. Instead, share sections incrementally.
Share Methods first. Your supervisor can catch experimental details you’ve missed or misremembered while memory is fresh.
Share Results next. Your supervisor can confirm your interpretation is accurate and suggest additional analysis if needed.
Share Introduction and Discussion once your supervisor has approved Results. These sections depend on Results being final.
Incremental review prevents massive rewrites late in the process.
Revision Rounds
Your supervisor will return a marked-up manuscript. Their comments may be contradictory, some wrong, some essential. Read all feedback, then decide what to implement. You are the expert on your data.
For rejected manuscripts that come back with reviewer comments, treat reviews as a gift. Reviewers have found genuine issues (even if their tone is harsh). Addressing these issues will make your manuscript stronger and more likely to be accepted at the next journal.
Responding to Reviewer Comments
When your manuscript is rejected and sent out for peer review, you’ll receive comments. Write a point-by-point response:
“Reviewer Comment: ‘Figure 2 lacks statistical analysis.’
Response: We have added p-values and error bars to Figure 2B (now Figure 2B-C) and updated the figure legend. We performed a two-way ANOVA… [explain the test]. This was not shown in the original submission because…”
Be respectful but defend your work when reviewers are wrong. Most editors appreciate thoughtful, evidence-based responses.
Next Steps
Once your manuscript is complete and approved by your supervisor, the next steps are:
Choosing a journal. Consider the journal’s scope, acceptance rate, and audience. If your work is novel with broad impact, target high-impact journals. If it’s solid but incremental, target specialized journals in your field. Check journal websites for author guidelines and target impact factor (if that matters to you or your field).
Writing a cover letter. Most journals require a brief cover letter explaining why your work is suitable for that journal. Keep it to one paragraph: “We report a novel mechanism by which X controls Y, which may inform treatment of disease Z. This manuscript has not been published elsewhere and is not under review at another journal.”
What to expect in peer review. Manuscripts are sent to 2-4 expert reviewers. Review takes 4-8 weeks. You’ll receive comments and an editor decision: accept (rare on first submission), minor revisions, major revisions, or reject. Most papers require at least one revision round. Respond to comments, resubmit, and repeat if necessary.
Your first manuscript is now ready to navigate peer review. Good luck.
If you want to go deeper on scientific writing craft, two books stand out: The Scientist’s Guide to Writing by Stephen Heard is the most practical book on scientific prose — it goes well beyond structure into how sentences and paragraphs actually work in scientific communication. Writing Science by Joshua Schimel focuses specifically on how to tell a compelling scientific story, which matters as much for grants as it does for papers.