Benchmarking integrative multiomics methods for disease classification and biomarker discovery

layout: true

<div class="my-footer"><span><a href="https://tonyliang19.github.io/ssrip_2023" style="color:white">tonyliang19.github.io/ssrip_2023</a></span></div>

---

background-image: url(materials/img/website_bkg.png)
background-position: top
background-size: contain

<br/><br/><br/><br/><br/><br/><br/><br/><br/><br/><br/>
.center[### Benchmarking integrative multiomics methods for disease classification and biomarker discovery]

.pull-left[
<br/>
####Tony Liang
#### Supervisor: Dr. Amrit Singh
####University of British Columbia

]

.pull-right[.center[.middle[
<br/>
<table>
<tr>
<td>![:scale 50%](materials/img/qr-slides.png)</td>
<td>![:scale 50%](materials/img/qr-code.png)</td>
<td>![:scale 50%](materials/img/qr-lab.png)</td>
</tr>
<tr>
<td><svg aria-hidden="true" role="img" viewBox="0 0 640 512" style="height:1em;width:1.25em;vertical-align:-0.125em;margin-left:auto;margin-right:auto;font-size:inherit;fill:currentColor;overflow:visible;position:relative;"><path d="M160 64c0-35.3 28.7-64 64-64H576c35.3 0 64 28.7 64 64V352c0 35.3-28.7 64-64 64H336.8c-11.8-25.5-29.9-47.5-52.4-64H384V320c0-17.7 14.3-32 32-32h64c17.7 0 32 14.3 32 32v32h64V64L224 64v49.1C205.2 102.2 183.3 96 160 96V64zm0 64a96 96 0 1 1 0 192 96 96 0 1 1 0-192zM133.3 352h53.3C260.3 352 320 411.7 320 485.3c0 14.7-11.9 26.7-26.7 26.7H26.7C11.9 512 0 500.1 0 485.3C0 411.7 59.7 352 133.3 352z"/></svg> Slides</td>
<td><svg aria-hidden="true" role="img" viewBox="0 0 640 512" style="height:1em;width:1.25em;vertical-align:-0.125em;margin-left:auto;margin-right:auto;font-size:inherit;fill:currentColor;overflow:visible;position:relative;"><path d="M392.8 1.2c-17-4.9-34.7 5-39.6 22l-128 448c-4.9 17 5 34.7 22 39.6s34.7-5 39.6-22l128-448c4.9-17-5-34.7-22-39.6zm80.6 120.1c-12.5 12.5-12.5 32.8 0 45.3L562.7 256l-89.4 89.4c-12.5 12.5-12.5 32.8 0 45.3s32.8 12.5 45.3 0l112-112c12.5-12.5 12.5-32.8 0-45.3l-112-112c-12.5-12.5-32.8-12.5-45.3 0zm-306.7 0c-12.5-12.5-32.8-12.5-45.3 0l-112 112c-12.5 12.5-12.5 32.8 0 45.3l112 112c12.5 12.5 32.8 12.5 45.3 0s12.5-32.8 0-45.3L77.3 256l89.4-89.4c12.5-12.5 12.5-32.8 0-45.3z"/></svg> Code</td>
<td><svg aria-hidden="true" role="img" viewBox="0 0 512 512" style="height:1em;width:1em;vertical-align:-0.125em;margin-left:auto;margin-right:auto;font-size:inherit;fill:currentColor;overflow:visible;position:relative;"><path d="M256 64C150 64 64 150 64 256s86 192 192 192c17.7 0 32 14.3 32 32s-14.3 32-32 32C114.6 512 0 397.4 0 256S114.6 0 256 0S512 114.6 512 256v32c0 53-43 96-96 96c-29.3 0-55.6-13.2-73.2-33.9C320 371.1 289.5 384 256 384c-70.7 0-128-57.3-128-128s57.3-128 128-128c27.9 0 53.7 8.9 74.7 24.1c5.7-5 13.1-8.1 21.3-8.1c17.7 0 32 14.3 32 32v80 32c0 17.7 14.3 32 32 32s32-14.3 32-32V256c0-106-86-192-192-192zm64 192a64 64 0 1 0 -128 0 64 64 0 1 0 128 0z"/></svg> Lab website</td>
</tr>
</table>
]]]

???

* Explain a little in the QR
* Thanks to compbio lab to allow this project, everyone helped!

---

# Land acknowledgement

.pull-left[
**I would like to acknowledge that I work on the traditional, ancestral, and unceded territory of the Coast Salish Peoples, including the territories of the xwməθkwəy̓əm (Musqueam), Skwxwú7mesh (Squamish), Stó:lō and Səl̓ílwətaʔ/Selilwitulh (Tsleil- Waututh) Nations.**

*Traditional*: Traditionally used and/or occupied by Musqueam people

*Ancestral*: Recognizes land that is handed down from generation to generation

*Unceded*: Refers to land that was not turned over to the Crown (government) by a treaty or other agreement

]

.pull-right[
.center[<img src ="materials/img/land_acknowledgement.gif" alt="matrix", width="80%"/>]
]

---

## Motivation

- Multiomics is a new emerging field of study

- Many methods, but which one is best at what case?

- Strategy of integrating multiomics data

- "Gap" between biologists and computer-based multiomics methods.

<!---

- Machine learning (`ML`) methods

- Reproducible workflows and parallel computation

- Lack of CS knowledge in generic biological field

- "Advanced" data science

- **Multiomics** problem could also be viewed as mathematical or statistical problem
--->

- **Multiomics** problem could also be viewed as mathematical or statistical problem

???

* Dont take too long in explain this
* Try making a 'consecutive' story

---

## Outline

.pull-left[
.large[Background]
 
 - **Multiomics**
 
 - **Data integration**

.large[Objectives]

.large[Methods]

.large[Expected Results]

.large[Conclusion ]

.large[Acknowledgements]

.large[.footnote[
For code to original repo see [.secondary[here]](https://github.com/tonyliang19/multiomics-pipeline)
]]
]

---

## Multiomics Data

Profiling of different molecular measurements (i.e. genes, proteins, metabolites) from same biological samples.

.pull-left[

- Compensate for missing or unreliable information in any single `omics` data

- Multiple source of evidence pointing to same biological process are more likely

<br>

**Omics** is the comprehensive study of all molecules of a particular type within an organism.

- Each `omics` data follow `$N \times P$` structure
  + `$N =$` Samples of Observations
  
  + `$P =$` Predictors/variables
  
  + High dimensional, `$N << P$`
]

.pull-right[
<br>

<br>

![:scale 100%](materials/img/multi-omics.png)
]

???

* Dont go into too much detail of math
* Use simple language

---

### Multiomics Data Integration

.center[
![:scale 90%](materials/img/types-methods.png)
]

???

* Left is typical example of bulk
* Right is most types of methods (there are more)

---

## Objectives

The objective is to benchmark methods for multiomics data integration by:

1. Curating publicly available multiomics data

2. Applying existing methods to simulated and real world datasets

3. Compare methods based on 
  + **classification accuracy **
  
  + **features selected**
  
  + **computation time**, etc

???

* Keep it short, since all details is explained in methods

---

.center[![:scale 77%](materials/img/workflow.png)]

???

* Explain, this is generic workflow of the pipeline

---

### Data

.pull-left-narrow[

- Simulated data

- Bulk data (.secondary[Majority])

- Single-cell data

- Spatial data
]

.pull-right-wide[
![](materials/img/omics.png)
]

Many of these data are **high dimensional**, quite abstract to illustrate in plain words.

???

Briefly explain what each consits of

Bulk -> majority of multiomics, this one ignores cell-type indenty, most basic one

Single-cell -> new single cell technologies like CITE-seq , scRNA-seq, scATAC-seq and ECCITE-seq

Spaital -> Spatial distirbution of cells is known

---

## Methods

- Using **Nextflow** that eases writing of data-intensive computational pipelines

- Running these pipelines on **UBC Advanced Research Computing cluster Sockeye**

.pull-left-narrow[
.secondary[Curate multiomics data from open-source databases
]

- Collect different omics data from same samples

- AnnData for `Python`

- MultiAssayExperiment for `R`

- Normalization , batch correction
]

.pull-right-wide[
.secondary[ `Benchmarking` integrative multiomics methods
]

- Evaluate series of methods based on their tasks with classic metrics

- Compare methods on different types of multiomics data

- Develop method that improve the predictive performance and
interpretability of the results.

- Use computational power and CS parallel algorithm with Nextflow and UBC ARC Sockeye
]

---

background-image: url(materials/img/nextflow-story.png)
background-size: contain
class: middle, center

<!---
.pull-left[
- Highly customizable

- Parallel computation

- Portable in multiple platforms
]
.pull-right[

- Community pipelines and support

- Any programming Language

- <svg aria-hidden="true" role="img" viewBox="0 0 512 512" style="height:1em;width:1em;vertical-align:-0.125em;margin-left:auto;margin-right:auto;font-size:inherit;fill:#e98a15;overflow:visible;position:relative;"><path d="M459.1 52.4L442.6 6.5C440.7 2.6 436.5 0 432.1 0s-8.5 2.6-10.4 6.5L405.2 52.4l-46 16.8c-4.3 1.6-7.3 5.9-7.2 10.4c0 4.5 3 8.7 7.2 10.2l45.7 16.8 16.8 45.8c1.5 4.4 5.8 7.5 10.4 7.5s8.9-3.1 10.4-7.5l16.5-45.8 45.7-16.8c4.2-1.5 7.2-5.7 7.2-10.2c0-4.6-3-8.9-7.2-10.4L459.1 52.4zm-132.4 53c-12.5-12.5-32.8-12.5-45.3 0l-2.9 2.9C256.5 100.3 232.7 96 208 96C93.1 96 0 189.1 0 304S93.1 512 208 512s208-93.1 208-208c0-24.7-4.3-48.5-12.2-70.5l2.9-2.9c12.5-12.5 12.5-32.8 0-45.3l-80-80zM200 192c-57.4 0-104 46.6-104 104v8c0 8.8-7.2 16-16 16s-16-7.2-16-16v-8c0-75.1 60.9-136 136-136h8c8.8 0 16 7.2 16 16s-7.2 16-16 16h-8z"/></svg> Requires some CS knowledge

]
--->

???

Only briefly mention the platform used and the bullets.
Rest could be explained in methods

---

## Expected Results

.large[ 1\. Availability of list of curated multiomics dataset]

<div class="datatables html-widget html-fill-item-overflow-hidden html-fill-item" id="htmlwidget-02880d8cbf9369760f4f" style="width:100%;height:auto;"></div>
<script type="application/json" data-for="htmlwidget-02880d8cbf9369760f4f">{"x":{"filter":"none","vertical":false,"fillContainer":false,"data":[["1","2","3","4","5","6","7","8","9","10"],["Multiomics modeling of the immunome, transcriptome, microbiome, proteome and metabolome adaptations during human pregnancy","","","","","","","Cytokine-induced molecular responses in airway smooth muscle cells inform genome-wide assocation studies of asthma","",""],["longitudinal","","","","","","","case-control","",""],["plasma","plasma","serum","swab","blood","plasma","plasma","primary airway smooth muscle cells","primary airway smooth muscle cells","primary airway smooth muscle cells"],["cfRNA","proteins","proteins","microbiota","cells","metabolites","proteins","genotypes","mRNA","methylation"],["RNAseq","Luminex","Luminex","DNAseq","CyTOF","mass spec","Somalogic","Illumina Omni2.5v8v1A/Human Core Array","Illumina Human HT-12 v4 array","Illumina Human Methylation EPIC Beadchip"],[37275,62,62,18548,354,3485,1300,null,18279,786326],[68,null,null,null,null,null,null,268,null,null],["4 timepoints","","","","","","","asthma/non-asthma","",""],["multinomial","","","","","","","binary","",""],["","","","","","","","age, smoking status, race, sex, treatment","",""],["Bioinformatics 2019 35(1):95-103.","","","","","","","Genome Med. 2020 12(1):64.","",""],["https://nalab.stanford.edu/multiomics-pregnancy/","","","","","","","to ask authors","GSE146374","GSE146376"]],"container":"<table class=\"display\">\n  <thead>\n    <tr>\n      <th> <\/th>\n      <th>Title<\/th>\n      <th>study design<\/th>\n      <th>sample-type<\/th>\n      <th>omic-type<\/th>\n      <th>technology<\/th>\n      <th>number of variables<\/th>\n      <th>number of observations<\/th>\n      <th>response<\/th>\n      <th>response-type<\/th>\n      <th>covariates<\/th>\n      <th>reference<\/th>\n      <th>data link/GEO<\/th>\n    <\/tr>\n  <\/thead>\n<\/table>","options":{"pageLength":2,"scrollX":true,"searching":false,"dom":"ftp","columnDefs":[{"className":"dt-right","targets":[6,7]},{"orderable":false,"targets":0}],"order":[],"autoWidth":false,"orderClasses":false,"lengthMenu":[2,10,25,50,100],"rowCallback":"function(row, data, displayNum, displayIndex, dataIndex) {\nvar value=data[1]; $(this.api().cell(row, 1).node()).css({'font-size':'85%'});\nvar value=data[2]; $(this.api().cell(row, 2).node()).css({'font-size':'85%'});\nvar value=data[3]; $(this.api().cell(row, 3).node()).css({'font-size':'85%'});\nvar value=data[4]; $(this.api().cell(row, 4).node()).css({'font-size':'85%'});\nvar value=data[5]; $(this.api().cell(row, 5).node()).css({'font-size':'85%'});\nvar value=data[6]; $(this.api().cell(row, 6).node()).css({'font-size':'85%'});\nvar value=data[7]; $(this.api().cell(row, 7).node()).css({'font-size':'85%'});\nvar value=data[8]; $(this.api().cell(row, 8).node()).css({'font-size':'85%'});\nvar value=data[9]; $(this.api().cell(row, 9).node()).css({'font-size':'85%'});\nvar value=data[10]; $(this.api().cell(row, 10).node()).css({'font-size':'85%'});\nvar value=data[11]; $(this.api().cell(row, 11).node()).css({'font-size':'85%'});\nvar value=data[12]; $(this.api().cell(row, 12).node()).css({'font-size':'85%'});\n}"}},"evals":["options.rowCallback"],"jsHooks":[]}</script>

.footnote[
This is a yet to be updated data table (under curation)
]

---

## Expected Results

.large[2\. Web portal app to quickly denote a method suitable to your custom data.]

- Time, error rates, number of features selected by method
  + Simulated (with/without noise)
  
  + Real dataset

---

### Limitations and Future Direction

.emphasis[
You don't need to understand tons of cs to use this project
]

---

## Conclusion

1. The best performing methods on the real-world data will provide insights into why certain methods are better than others with respect predictive performance and capturing the correct underlying biological processes.

2. Inform the development of a new integrative method that is applicable to different types of multiomics data.

3. Highly relevant to the community utilizing multiomics data for
their research.
  + Ease of use
  + Collaborations
  + Development of new tool

---

class: middle
background-image: url(materials/img/website_bkg.png)
background-position: top
background-size: contain

.large[.secondary[Thanks to my lab, the [CompBio Lab](https://cbl-hli.med.ubc.ca/)]]
.pull-left[
- Dr. Amrit Singh
- Dr. Dantong Zhu
- Dr. Young Woong Kim
- Dr. Maryam Ahmadzadeh
- Michael Yoon
]
.pull-right[
- Jeffrey Tang
- Rishika Daswani
- Roy He
- Samuel Leung
- Akshdeep Sandhu
- Asees Singh
]

<br>

And acknowledging to UBC ARC, work learn,  and Nextflow for enabling the successful execution of  this project.

.large[
.footnote[
For more information regarding Nextflow, please check [here](https://www.nextflow.io/)
]]

<!----
Credits to:

[Stat 406 Lecture Slides @DJM](https://ubc-stat.github.io/stat-406/)