Protein Structure

Lecture 11/12 Video 4

Quiz ChatGPT conversation

🧪 MS Lecture 3 (Part 4) — Bottom-Up Proteomics & Data Acquisition

🔬 1. Big Picture: What are we doing?

This lecture focuses on how we select and analyze molecules in mass spectrometry (MS), especially in:

Bottom-up proteomics → proteins are digested into peptides, then analyzed
MS1 → MS2 workflow → detect ions first, then fragment them for identification

The key question: 👉 Out of thousands of signals, which ones do we analyze further?

📊 2. What is a “Feature” in MS?

In real MS data, you don’t just see one peak per molecule.

Instead, each molecule appears as a multi-dimensional feature:

m/z axis (mass-to-charge) → isotope peaks (+1, +2, +3, etc.)
Retention time (chromatography) → when it elutes

So a feature = 👉 all isotope peaks + their time evolution

📌 Important:

A single molecule = many peaks over time
This makes data complex and crowded

Real datasets look like: 👉 thousands of overlapping features (very “busy”)

🎣 3. Untargeted (Discovery) Proteomics

Goal:

👉 Identify everything in the sample

Characteristics:

No prior assumptions
“Fishing expedition”
Huge datasets

⚠️ Challenge:

Data is so complex that you depend heavily on software
Requires:
- Bioinformatics
- Database matching
- High computing power

⚙️ 4. Data-Dependent Acquisition (DDA)

💡 Core idea:

👉 Let the instrument decide what to fragment based on MS1

Workflow:

Perform MS1 scan
Select top N most intense peaks
Fragment them (MS2)
Repeat

⏱️ Important concept: Duty Cycle

Duty cycle = total time for one loop: MS1 → select → MS2 (multiple times) → next MS1

Trade-off:

Faster = more scans
Slower = better signal

🔋 Fill Time

Time used to accumulate ions before measurement.

Trade-off:

Short → fast but noisy
Long → better signal but slower

🚫 Limitation of DDA

👉 Bias toward high-intensity ions

Meaning:

Low-abundance peptides may be missed
You don’t get full coverage of the proteome

🌐 5. Data-Independent Acquisition (DIA)

💡 Core idea:

👉 Don’t pick specific ions — fragment everything in a range

How it works:

Instead of selecting peaks:

Divide m/z into windows (e.g., 10–100 m/z wide)
Fragment all ions in each window

🧠 Key difference vs DDA:

Feature	DDA	DIA
Selection	Top intense peaks	All ions in window
Bias	Yes	No
Data complexity	Lower	Much higher
Coverage	Limited	Broad

⚠️ Challenge in DIA:

MS2 spectra become: 👉 mixtures of fragments from multiple peptides

So:

Harder to interpret
Requires advanced computation

✅ Advantage:

👉 You capture low-abundance peptides too

🔁 6. Combining DDA and DIA

In practice, experiments often combine both:

DDA first
- Identify peptides
- Build reference database
DIA next
- Quantify those peptides more accurately

👉 This improves both identification + quantification

🧩 7. Real MS2 Data is Incomplete

In theory:

You expect full b-ion and y-ion series

In reality:

Only partial fragments
Noise present
Missing peaks

Result:

👉 Matching spectra becomes difficult

You compare:

Theoretical spectrum (expected)
Experimental spectrum (observed)

And they rarely match perfectly

🧮 8. Peptide Identification & Scoring

Software (e.g., MaxQuant, Andromeda) assigns scores based on:

Number of matching peaks
Peak intensities
Fragment patterns

⚠️ Problem:

Two peptides can get same score

🧠 How do we decide?

We use biological assumptions:

Example:

Peptide A: simple, expected digestion → likely correct
Peptide B: requires multiple rare modifications → unlikely

👉 Choose the most plausible explanation

🧬 9. Protein Inference Problem

Problem:

👉 A peptide can belong to multiple proteins

Example:

Peptides:

1, 2, 3

Proteins:

A → contains 1,2,3
B → contains 1,3
C → contains 2

Conclusion:

👉 You cannot uniquely assign proteins

Solution: Protein Groups

Group proteins together
Choose “lead protein” (most likely)
But acknowledge uncertainty

🔑 Key concept:

Unique peptides → strong evidence
Shared peptides → ambiguous

🎯 10. Targeted Proteomics

Goal:

👉 Focus on specific proteins

Instead of “find everything”, you ask:

Is protein X present?
How much is it?

How to reduce complexity:

🧪 Sample-level:

Fractionation
Enrichment (e.g., phosphoproteins)
Isolate specific protein types

⚙️ MS-level:

Use targeted acquisition (e.g., MRM)

🚀 Benefits:

Less data
Faster analysis
Better quantification
Higher sensitivity

📌 11. Key Takeaways

🔑 Core ideas:

MS data is multi-dimensional and complex
Selecting ions is a major challenge
Two main strategies:
- DDA → selective, biased
- DIA → comprehensive, complex
Data interpretation relies heavily on:
- Software
- Statistical scoring
- Biological assumptions
Protein identification is not direct
- Requires inference from peptides

🧠 Conceptual Summary

Think of proteomics like:

🎣 DDA = fishing with a net that catches only big fish 🌊 DIA = sweeping the entire ocean (but sorting is harder)

Quiz

Score: 0/31 (0%)

Q0. In bottom-up proteomics, why is precursor selection after MS1 an important problem?

Because MS1 data contains many overlapping features across m/z and retention time, so the instrument must decide which ions to fragment

Because MS1 always identifies the complete protein sequence directly

Because MS2 spectra cannot be produced from peptide ions

Because isotope patterns are removed before precursor selection

Q1. What does an MS feature typically include?

Only one monoisotopic peak at one retention time

A group of isotope peaks belonging to an ion species, observed over a retention-time profile

Only the fragment ions produced after collision-induced dissociation

Only proteins that have already been identified by database searching

Q2. What is the main goal of untargeted or discovery proteomics?

To measure only one known protein with maximum sensitivity

To identify as many proteins or peptides as possible without knowing in advance what will be present

To avoid bioinformatics by manually reading every spectrum

To fragment only phosphorylated peptides

Q3. In data-dependent acquisition, DDA, how are precursor ions usually selected for MS2?

All ions within fixed m/z windows are fragmented together

The lowest-intensity ions are selected first

A set of intense precursor ions from the MS1 scan is selected and fragmented sequentially

Only proteins with known post-translational modifications are selected

Q4. What is the duty cycle in DDA?

The total time needed to perform an MS1 scan, select precursors, acquire MS2 spectra, and return to the next MS1 scan

The time required for trypsin to digest a protein sample

The amount of time a peptide spends in the chromatography column

The time needed to search a database after the experiment

Q5. Why can DDA miss low-abundance peptides?

Because DDA never performs MS1 scans

Because DDA usually prioritizes high-intensity precursor ions, leaving weaker signals unfragmented

Because low-abundance peptides cannot be ionized by soft ionization

Because low-abundance peptides do not have isotope patterns

Q6. What is the purpose of dynamic exclusion in DDA?

To permanently remove all isotope peaks from MS1 spectra

To prevent recently fragmented precursor ions from being immediately selected again, allowing other ions to be sampled

To remove all low-abundance peptides from the dataset

To make MS2 spectra independent of MS1 scans

Q7. How does DIA differ from DDA in precursor selection?

DIA fragments only one manually selected precursor ion per experiment

DIA skips MS2 entirely

DIA fragments all ions within predefined m/z windows rather than selecting individual high-intensity precursors

DIA only works for purified single proteins

Q8. Why are DIA MS2 spectra often more complex than DDA MS2 spectra?

Because DIA contains fragments from multiple precursor ions within the same m/z window

Because DIA removes all fragment ions below the precursor m/z

Because DIA cannot use chromatography

Because DIA only fragments the monoisotopic peak

Q9. What is a major advantage of DIA compared with DDA?

It requires no computational interpretation

It can sample ions more comprehensively, including lower-abundance precursors that DDA may miss

It always produces simpler MS2 spectra

It identifies intact proteins directly without peptide digestion

Q10. Why might a long-gradient DDA experiment be performed before DIA analysis?

To create a reference set of identified peptide ions that can later help interpret or quantify DIA data

To destroy low-abundance peptide ions before DIA

To remove the need for chromatography in DIA

To convert proteins into intact ions without digestion

Q11. Why is peptide-spectrum matching not always straightforward?

Because experimental MS2 spectra often contain incomplete fragment-ion series, missing peaks, unexpected peaks, and noise

Because theoretical spectra always perfectly match experimental spectra

Because CID produces no b- or y-ions

Because every peptide has a unique mass and therefore needs no MS2 evidence

Q12. In peptide identification, why may software use assumptions in addition to scoring?

Because two peptide candidates can receive similar or equal scores, so biological plausibility helps choose the more likely identification

Because assumptions are more important than experimental data

Because all peptide candidates are equally correct

Because database searches do not use fragment ions

Q13. Which peptide identification would generally be considered more plausible after trypsin digestion?

A peptide ending in lysine or arginine with no missed cleavages or unexpected modifications

A peptide requiring multiple rare modifications and several missed trypsin cleavage sites

A peptide with no fragment-ion matches

A peptide that cannot exist in the searched protein database

Q14. What is protein inference in bottom-up proteomics?

The process of deciding which proteins are likely present based on peptide-level evidence

The direct sequencing of intact proteins without digestion

The conversion of proteins into isotope peaks

The exclusion of all shared peptides from MS1 scans

Q15. A feature in LC-MS includes information across both m/z and retention time.

True

False

Q16. Untargeted proteomics is usually hypothesis-driven and only measures preselected proteins.

True

False

Q17. DDA usually fragments precursor ions sequentially, one precursor at a time.

True

False

Q18. The duty cycle is unrelated to how many MS2 spectra can be collected during an LC-MS run.

True

False

Q19. Fill time involves allowing ions to accumulate so there is enough signal for measurement.

True

False

Q20. DDA is biased toward intense precursor ions and may miss weaker signals.

True

False

Q21. DIA relies on selecting only the top few most intense MS1 peaks for fragmentation.

True

False

Q22. In DIA, wider isolation windows can increase the complexity of MS2 spectra.

True

False

Q23. DIA data interpretation often requires substantial computational analysis.

True

False

Q24. Experimental MS2 spectra usually contain complete b-ion and y-ion series for every peptide.

True

False

Q25. Peptide scoring compares experimental spectra with theoretical spectra generated from candidate peptide sequences.

True

False

Q26. A peptide candidate requiring many unlikely modifications and missed cleavages may be considered less plausible than a simpler candidate with the same score.

True

False

Q27. Protein inference is unnecessary in bottom-up proteomics because every peptide uniquely identifies exactly one protein.

True

False

Q28. A unique peptide provides stronger evidence for a specific protein than a peptide shared by several proteins.

True

False

Q29. Protein groups are used when peptide evidence cannot clearly distinguish between several possible proteins.

True

False

Q30. Targeted proteomics focuses on specific proteins or peptides and can reduce data complexity compared with discovery proteomics.

True

False