AGENT SKILL

thesis-structure

Use when planning or checking the architecture of a PhD / MPhil thesis in the life sciences or bioinformatics — monograph vs article-based, chapter sequence, IMRaD-based chapters, abstract-to-conclusion flow, and the examiner expectations that the structure must meet.

License 许可
MIT
最近更新
2026-10-30
Stars
★ 8
Forks
2

01 前言 Front Matter

name
thesis-structure
license
MIT
description
Use when planning or checking the architecture of a PhD / MPhil thesis in the life sciences or bioinformatics — monograph vs article-based, chapter sequence, IMRaD-based chapters, abstract-to-conclusion flow, and the examiner expectations that the structure must meet.

03 SKILL.md 正文

<!-- metadata: category: thesis-writing oversight: medium version: 1.0.0 author: Pradyumna Jayaram tags: [thesis-writing, structure, imrad, monograph] difficulty: intermediate -->

PhD Thesis Structure (Life Sciences & Bioinformatics)

The structural spine of a thesis before a word is written. Choosing between a monograph (original chapters) and an article-based thesis (submitted papers with an integrative frame) is the first decision, and it fixes everything downstream: chapter order, the role of the Introduction, and what the examiner will expect to find at each position.

Hard rules

  1. No fabricated citations. Every in-text citation you list in the structure check must resolve to a verifiable record (DOI, PMID, arXiv ID, ISBN, or URL with a retrieval date). A structure that asserts "chapter 3 builds on (Kumar 2019)" when that record does not resolve is a defect, not a placeholder.
  2. No claim without provenance. Every structural claim about your thesis (this chapter is chapter 4, it contains n=X results, it follows chapter 3) must point to a section, heading, or figure the agent can open and confirm.
  3. No silent failure. Every tool that scans or outlines the draft (a heading extractor, a section-counting pass, a word-count) must report what happened on non-zero. A "your draft has no methods section" that came from a parse error, not the document, is a false finding — flag the parse, not the thesis.
  4. Decide monograph vs article-based first, and say so. The two have different examiner expectations; an unlabeled hybrid confuses the reviewer.

When to use

  • Before writing, to set the chapter sequence and fix the monograph / article-based choice.
  • Mid-thesis, to check that each chapter has the IMRaD shape examiners expect.
  • At the "does this read as a thesis, or as a pile of papers?" review.

When NOT to use

  • The reproducibility of methods and code — use thesis-methods-reproducibility.
  • The viva / defense — use thesis-defense-viva.
  • The single-particle of an Introduction or Abstract paragraph — use manuscript-structure.
  • A journal paper, not a thesis — use the scientific-writing skills (manuscript-structure, cover-letter).

Prerequisites

  • The supervisor has confirmed the thesis type the department expects (monograph is the default in many Indian and UK programs; article-based needs explicit permission in others).
  • A rough inventory of completed work: which experiments / datasets / papers exist, and in what order they were done.

The two archetypes

Monograph Article-based
Core Original chapters written by you 2–3 submitted/published papers + an integrative frame
Ch 1 Literature review (you own the narrative) Intro that integrates the papers into one argument
Ch 2..n One theme/organism/method each Each paper becomes a chapter (often verbatim, with permission)
Final Discussion + general conclusion General discussion reconciling all papers
Examiner risk Coherence: one argument across chapters Redundancy: repeated intros/methods; need one unifying thread

The IMRaD spine

Most life-science chapters are IMRaD (Introduction–Methods–Results– Discussion). For a whole thesis, map that onto chapters:

  1. Chapter 1 — Introduction / Background. The whole-thesis "why now." Build the gap, not a survey. (This is the thesis-scale version of a paper intro; keep it ~8–15% of the thesis.)
  2. Chapter 2 — (thesis-scale) Methods & resources, or fold into each chapter. In monographs a shared-methods chapter de-duplicates; in article-based the methods live inside each paper chapter.
  3. Chapters 3..n — Results chapters, one theme each, IMRaD internally.
  4. Final chapter — General discussion + conclusion. Synthesize across chapters; state the contribution and the limits; name the follow-ups.

An abstract at the very front must contain: context → the gap → what you did → the main finding → the contribution, in 250–350 words (or the program's limit).

Structure workflow

  1. Choose the archetype and record it in the draft's front matter so the examiner sees the intent.
  2. Draft the chapter map — one bullet per chapter with a single-sentence "this chapter establishes ___." If a chapter's sentence is empty, cut it.
  3. Assign every result to exactly one chapter. No result sits in two; no chapter is empty. (A shared methods chapter is the one legitimate exception.)
  4. Check IMRaD internally per results chapter: does it have intro (what this chapter does), methods (how), results (what), discussion (so what)?
  5. Front + back matter: title page, abstract, list of figures, list of tables, abbreviations, acknowledgements, then the body, then references, then appendices. Order and presence are checked mechanically, so run a heading scan.
  6. Length sanity-check against the program's word/page cap (many programs cap ~60–80 pages of body text). A thesis that is too long is a structure problem, not a writing problem.

Pitfalls

  • The unlabeled hybrid — some monograph chapters, some paper chapters, and no statement of which it is. Examiners read the mess as "I did not think about the structure."
  • Article-based redundancy — three near-identical introductions; no integrative thread. The frame chapter is what makes it a thesis, not a folder of PDFs.
  • Results in two chapters — the same figure appears in ch. 4 and ch. 6; the examiner is left to work out which is canonical.
  • A Discussion that re-lists Results instead of interpreting across chapters.
  • Missing front/back matter (no list of figures, no abbreviations) — small, but read as sloppiness on day one.
  • Too long — padding dilutes the contribution; trim to one argument.

Validation

  • The archetype (monograph / article-based) is stated on the title/foreword.
  • A heading scan confirms the full front → body → back matter sequence, with no duplicate results chapters.
  • Each results chapter passes the internal IMRaD check.
  • The abstract is within the program's word cap and carries all five beats.
  • Every citation named in the structure check resolves.

Cross-references

  • thesis-methods-reproducibility — the methods, code, and figure discipline the structure must support.
  • thesis-defense-viva — what the examiner will probe at the boundary of each chapter.
  • manuscript-structure — the IMRaD / structure models for a single paper, which a thesis chapter reuses at sentence and paragraph scale.

References

  • IMRaD and the five standard structure models: local academic-research-skills-main/academic-paper/references/paper_structure_patterns.md.
  • Article-based thesis guidance: bioinformatics-thesis-main/thesis-writing/SKILL.md and thesis-writing-skills-main (local reference collections).
  • Program-specific page caps and monograph-vs-article rules live in your department's thesis manual — the authoritative, local source.

Changelog

  • 1.0.0 (2026-06-18) — Initial release. Monograph vs article-based decision, the IMRaD chapter spine, front/back-matter sequence, and the examiner-expectation checks that structure must satisfy.

04 同仓库更多 Skills