index.utf8

class: primary

p.caption {
  font-size: 0.6em;
}

.large { font-size: 200% }

.medium-large { font-size: 130% }

.small{ font-size: 80% }

.tiny{ font-size: 40% }

.center-left {
  position:          relative;
  top:               50%;
  transform:         translateY(50%);
}
.center-right {
  position:          relative;
  top:               50%;
  transform:         translateY(10%);
}

.remark-slide-content {
  background-color: #FFFFFF;
  border-top: 80px solid #2b0a5e;
  font-size: 28px;
  font-weight: 300;
  line-height: 1.5;
  padding: .5em 1em .5em 1em
}

.inverse {
  background-color: #2b0a5e;
  text-shadow: none;
}

.right-column {
	color: #000000;
	width: 30%;
	height: 92%;
	float: right;
}

.left-column {
  width: 68%;
  float: left;
}

.remark-slide-number {
	display: none;
}

.remark-slide-content:after {
    content: "";
    position: absolute;
    bottom: 0px;
	left: 20px;
    height: 60px;
    width: 400px;
	font-size: 12px;
    background-repeat: no-repeat;
    background-size: contain;
	background-image: url("img/The-Ohio-State-University-Wexner-Medical-Center.png")
}

</style>

<h2 style=font-size:45px> Improving pathway analysis of
lipidomic and metabolomic data through comprehensive functional
annotation and network approaches </style>

.pull-left[
<h6 style=font-size:20px>
<br>
Andrew Patt, Doctoral Candidate
<br>
<br>
National Center for Advancing Translational Science/The Ohio State University
</style>
]

.pull-right[
.center[
<img src="img/network_example.png" width="250px" />
]
]

---

# Relational database of Metabolic Pathways (RaMP)
.pull-left[
.small[
- Multiomic pathway mySQL database integrating
information from **KEGG**, **HMDB**, **WikiPathways** and **Reactome**

- Currently contains pathway information for > 13,000 metabolites and >14,000 
  transcripts, totalling 51,526 pathways and 536,245 associations

- Associated R package allows for easy querying of database, as well as pathway analysis of
metabolite and transcript data individually or concurrently 
</br>
</br>
]

.tiny[
[Zhang B, Hu S, Baskin E, Patt A, Siddiqui JK, Mathé EA. RaMP: A Comprehensive Relational Database of Metabolomics Pathways for Pathway Enrichment Analysis of Genes and Metabolites. Metabolites. 2018;8(1):16. Published 2018 Feb 22. doi:10.3390/metabo8010016](https://pubmed.ncbi.nlm.nih.gov/29470400/)
]]

.pull-right[
</br>
<img src="img/RaMP_scheme.jpg" width="500px" />

.tiny[
</br>
</br>
[Patt A, Siddiqui J, Zhang B, Mathé E. Integration of Metabolomics and Transcriptomics to Identify Gene-Metabolite Relationships Specific to Phenotype. Methods Mol Biol. 2019;1928:441-468. doi:10.1007/978-1-4939-9027-6_23](https://pubmed.ncbi.nlm.nih.gov/30725469/)
]]

---

# Planned RaMP expansions

.pull-left[
- **Metabolic pathways:**
  - HumanCyc (255+ pathways)
  - Pathbank (78,488 pathways)
]

.pull-right[
- **Chemical Structures:**
  - InChIKeys
  - LyChIKeys
]

--
.tiny[
</br>
]

| DB          | # lipids | Chemical class | Subcellular location | Structure | Reactions | Function |
|-------------|----------|----------------|----------------------|-----------|-----------|----------|
| LION/Web    | > 50,000 | No             | Yes                  | No        | No        | Yes         |
| SwissLipids | 777,657  | Yes            | Yes                  | Yes       | Yes       | No         |
| Lipid Maps  | 43,636   | Yes            | No                   | Yes       | No        | No         |
| LipidPedia  | 4,487    | No             | Yes                  | No        | Yes       |  Yes        |

---

# Metabolite Similarity Networks

.pull-left[
.small[
- Through RaMP, we are collecting data that describes relationships between metabolites:
  - Shared pathway activity
  - Mutual participation in reactions
  - Chemical structural similarity
  - Associations with disease, etc...

- Using this information, we can build knowledge networks describing the information landscape of the database
] ]

.pull-right[

<img src="img/small_network_example.png" width="600px" />
]

---

# Challenges in metabolite/lipid pathway analysis
.pull-left[
.small[
- Lack of pathway annotations

- Incorporating biological and chemical annotations offer better coverage

- Conventional pathway overrepresentation analysis does not account for redundancy of pathway annotations
	
	- Pathway annotations are not independent, which is an assumption of the Fisher's/Hypergeometric tests

]
]

.pull-right[
<img src="img/pathway_coverage_2.png" width="800px" />
]

---

# Metabolite Structure, Pathway and Annotation Networks
.center[
<img src="img/network_schematic1.png" width="900px" />
]
---

# Metabolite Structure, Pathway and Annotation Networks
.center[
<img src="img/network_schematic2.png" width="900px" />
]
---

# Metabolite Structure, Pathway and Annotation Networks
.center[
<img src="img/network_schematic3.png" width="900px" />
]
---

# Metabolite Structure, Pathway and Annotation Networks
.center[
<img src="img/network_schematic4.png" width="900px" />
]
---

# Topological analysis/Enrichment analysis

.pull-left[
.small[
1. Run random walks with restarts algorithm using metabolites of interest as seed node set

2. Run random walks with restarts using random node set many times

3. Calculate percentile of true score  in random score distribution by node

4. Filter network down to highly similar nodes to seed set

5. Identify clusters of related metabolites in final model and perform
   enrichment analysis
]
]

.pull-right[
<img src="img/network_topology2.png" width="900px" />

]

---

# Example study: Metabolomics of liposarcoma

.pull-left[
.small[
- Three treatment-responsive (MDM2-low) liposarcoma cell lines were
  compared to three unresponsive (MDM2 high) cell lines using the
  Metabolon platform

- Statistical analysis identified 18 metabolites different between the
  groups, 10 of which mapped to pathways in the KEGG database

- Major differences were observed in lipids including ceramides and fatty acids

]

</br>

.tiny[
[Patt A, Demoret B, Stets C, et al. MDM2-Dependent Rewiring of Metabolomic and Lipidomic Profiles in Dedifferentiated Liposarcoma Models. Cancers (Basel). 2020;12(8):E2157. Published 2020 Aug 4. doi:10.3390/cancers12082157](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7463633/)
]
]

.pull-right[
![](index_files/figure-html/unnamed-chunk-11-1.png)
]

---

# Network Method Preliminary Results in Liposarcoma
.center[
<img src="img/Network_figure_2.png" width="900px" />
]
---

# Advantages/Disadvantages of Method

.pull-left[
**PROS** 
- Avoids issues arising from pathway interdependence by accounting
  for overlap in network modeling
  
- Improved coverage of annotations by incorporating sources other than pathways
]

.pull-right[
**CONS**
- Enriching by module may enrich spurious relationships

- More computationally intensive
]

---

# Future Extensions

- Optimizations:

- Incorporate KS test for improved statistical validity

- Identify best network topology analysis method

- Identify best network fusion approach

- Demonstrate improved reproducibility

- Build networks using lipid annotations

---

# Acknowledgements

.pull-left[
.small[
**Mathé Lab**
- Dr. Ewy Mathé, OSU/NCATS
- Tara Eicher, OSU/NCATS
- Kevin Ying, OSU
- Dr. Garrett Kinnebrew, OSU
- Kyle Spencer, OSU/NCATS

**Collaborators/Advisors**
- Dr. James Chen, OSU
- Dr. Kevin Coombes, OSU
- Dr. Zachary Abrams, OSU
- Dr. Lang Li, OSU

]]

.pull-right[
.small[
**Collaborators/Advisors**
- Dr. Rachel Kopec, OSU
- Dr. Tim Garrett, Florida State University
- Dr. Jeremy Koelmel Florida State University
- John Braisted, NCATS
- NCATS informatics core

**Funding from**
- The OSU Clinical and Translational Research Informatics Training
Program (4T15LM011270-05)
- Systems and Integrative Biology training program (T32GM068412) ]]

---

# Questions?

.pull-left[
<iframe src='https://gfycat.com/ifr/LongLateAcouchi' frameborder='0' scrolling='no' allowfullscreen width='480' height='390'></iframe><p><a href="https://gfycat.com/discover/question-gifs">from Question GIFs</a> <a href="https://gfycat.com/longlateacouchi-question-mark-confused-what-huh">via Gfycat</a></p>
]

.pull-right[
- Slides link: https://andyptt21.github.io/MANA_2020

- RaMP link: https://rampdb.ncats.io/
]