This document provides an introduction to proteomics. It discusses why proteomics is important as proteins carry out most biological functions. It also mentions valuable publicly available proteomics resources and applications of machine learning in proteomics data analysis and interpretation. The document describes how proteomics data is generated using trypsin digestion and mass spectrometry techniques. It discusses the different dimensions of proteomics data and how experiments are designed using labeling or tagging approaches. Finally, it outlines the features of proteomic data, approaches to data analysis including de novo sequencing and identifying post-translational modifications, relevant bioinformatics tools and publicly available cancer proteomics resources.

1. Introduction to proteomics

2. Why proteomics?
● Most of the stuff are done by proteins
● Valuable publicly available resources
● Post translational modifications may influence epigenetics
● Many recent applications of machine learning in
○ Preprocessing raw files
○ Analysis
○ Interpreting
● Proteogenomics
○ Proteoepigenomics?

3. How does a peptide look like

4. How does a peptide look like

5. How is proteomics data generated?
Image from: http://www.premierbiosoft.com/tech_notes/mass-spectrometry.html

6. Using trypsin to limit possibilities
Trypsin cleaves peptide chains mainly at the carboxyl side of the amino acids
lysine or arginine, except when either is followed by proline.

7. 1st dimension of data: time

8. Other dimensions
Slide from Andrei Drabovich

9. Possible precursor fragmentation
Slide from Thomas Kislinger

10. In summary
Slide from Thomas Kislinger

11. How are proteomics
experiments designed?
A. Labeling Cysteine after
obtaining peptide mixture
B. Labeling with heavy amino
acids while growing cells or
animal
C. Reaction with iTRAQ tags
Slide from Thomas Kislinger

12. What are the features for each peptide?
1. Time of elution
2. m/z and intensity of precursor ion (trypsinated peptides)
3. m/z and intensity of fragmented ions (mono-oligo peptides) from each
precursor ion

13. How to analyze the data?
1. Possible expectations from a FASTA file
2. Publicly available data from known proteins
3. De novo sequencing:
a. What are the possible peptides for a given precursion ion m/z
b. Which of these peptides are better supported with m/z of obtained b/y ions?

14. How to analyze the data?
Zolg, Daniel P., et al. "Building ProteomeTools based on a complete synthetic human proteome." Nature methods 14.3 (2017): 259.

15. How to infer post-translational modifications
Chick, Joel M., et al. "A mass-tolerant database search identifies a large proportion of unassigned spectra in shotgun proteomics as modified peptides."
Nature biotechnology 33.7 (2015): 743.

16. Bioconductor
packages

17. Publicly available cancer proteomics resources
● The Clinical Proteomic Tumor Analysis Consortium (CPTAC)
○ Breast cancer
○ Ovarian cancer
○ Colorectal cancer
○ Rectal adenocarcinoma
○ Skin carcinoma
○ Oral squamous cell carcinoma
○ Treated and resistant triple negative breast cancer xenografts