Skip to content

Commit

Permalink
Beijing talk
Browse files Browse the repository at this point in the history
  • Loading branch information
@gavryushkin
gavryushkin committed Jul 5, 2024
1 parent 63b1416 commit 969eaad
Showing 1 changed file with 337 additions and 0 deletions.
337 changes: 337 additions & 0 deletions talks/2024_Beijing_Alex.html
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,337 @@
<!doctype html>
<html lang="en">

<head>
<meta charset="utf-8">
<title>
Algorithms for precision oncology - Alex Gavryushkin
</title>
<meta name="author" content="Alex Gavryushkin">
<meta name="viewport" content="width=device-width, initial-scale=1.0, maximum-scale=1.0, user-scalable=no">
<meta name="apple-mobile-web-app-capable" content="yes" />
<meta name="apple-mobile-web-app-status-bar-style" content="black-translucent" />
<link rel="stylesheet" href="revealjs/dist/reveal.css">
<link rel="stylesheet" href="revealjs_themes_biods/simple_gavruskin.css" id="theme">
<link rel="stylesheet" href="revealjs/plugin/highlight/zenburn.css">
<link rel="icon" href="../favicon.ico" />
</head>

<body>
<div class="reveal">
<div class="slides">

<section>
<h3>
Algorithms for reconstructing within-patient tumor histories to support precision oncology
</h3>
<h4>
<a href="https://biods.org/alex/">Alex Gavryushkin</a>
</h4>
<h5>
<a href="https://biods.org/people/">
<img data-src="../assets/bioDS_lab_UC_logo.png" width=20%>
</a>
</h5>
<font size="5">
IVPP (Chinese Academy of Sciences), July 7, 2024, Beijing
</font>
</section>

<section>
<h3>Disclaimers</h3>
</section>

<section data-background-image="./images/what_is_cancer_evo_problem.svg">
</section>

<section>
<h4>Precision oncology</h4>
<img data-src="../assets/precision_oncology_by_vicka.svg">
</section>

<section>
<h4>Phylogenetic trees on tumor cells</h4>
<img data-src="./images/tumor_cell_tree.svg" width=80%>
</section>

<section data-background-image="./images/multiomic_cancer_evo.svg"
data-background-size=80%>
</section>

<section data-background-image="./images/spatial_phylo_on_all_cells.svg"
data-background-size=80%>
</section>

<section data-background-image="./images/pathologist_annotated_cells.svg"
data-background-size=80%>
</section>

<section data-background-image="./images/twenty_quality_cells.svg"
data-background-size=80%>
</section>

<section data-background-image="./images/cells_with_same_genotypes.svg"
data-background-size=80%>
</section>

<section data-background-image="./images/tree_on_all_quality_cells_expression.svg"
data-background-size=80%>
</section>

<section data-background-image="./images/tree_on_cells_with_same_genotype_expression.svg"
data-background-size=80%>
</section>

<section data-background-image="./images/tree_on_cells_with_same_genotype_snv.svg"
data-background-size=80%>
</section>

<section data-background-image="./images/tree_on_cells_with_same_genotype_expression_snv.svg"
data-background-size=80%>
</section>

<section>
<h3>Problem</h3>
<h4>This result is not consistent across datasets &mdash; why?</h4>
</section>

<section data-background-image="./images/visium_patient_one.svg"
data-background-size=70%>
</section>

<section data-background-image="./images/visium_patient_two.svg"
data-background-size=70%>
</section>

<section data-background-image="./images/visium_challanges.svg"
data-background-size=70%>
</section>

<section data-background-image="./images/visium_challanges_we_tackle.svg"
data-background-size=70%>
</section>

<section data-background-image="./images/expression_data_discretization.svg"
data-background-size=70%>
</section>

<section data-background-image="./images/gene_filtering_options.svg"
data-background-size=70%>
</section>

<section data-background-image="./images/patient_two_maxlik.svg"
data-background-size=50%>
</section>

<section data-background-image="./images/patient_one_maxlik.svg"
data-background-size=50%>
</section>

<section data-background-image="./images/visium_pathologist_annotation_both_patients.svg"
data-background-size=70%>
</section>

<section data-background-image="./images/visium_expected_clades.svg"
data-background-size=60%>
</section>

<section data-background-image="./images/visium_manually_selected_cells.svg"
data-background-size=60%>
</section>

<section data-background-image="./images/visium_patien1_bayesian_tree.svg"
data-background-size=60%>
</section>

<section data-background-image="./images/visium_p1_highest_counts.svg"
data-background-size=70%>
</section>

<section data-background-image="./images/visium_p1_highest_counts_bayesian_tree.svg"
data-background-size=70%>
</section>

<section data-background-image="./images/visium_p1_similar_counts.svg"
data-background-size=70%>
</section>

<section data-background-image="./images/visium_p1_similar_counts_bayesian_tree.svg"
data-background-size=70%>
</section>

<section data-background-image="./images/visium_p2_similar_counts.svg"
data-background-size=70%>
</section>

<section data-background-image="./images/visium_p2_similar_counts_bayesian_tree.svg"
data-background-size=70%>
</section>

<section data-background-image="./images/visium_external_branches_effect.svg"
data-background-size=70%>
</section>

<section data-background-image="./images/visium_number_effected_clades_effect.svg"
data-background-size=40%>
</section>

<section data-background-image="./images/visium_phylo_distance_total_counts.svg"
data-background-size=70%>
</section>

<section data-background-image="./images/visium_conclusion.svg"
data-background-size=60%>
</section>

<section>
Finding important genes
</section>

<section>
<h4>Model-based approaches to genotype-phenotype data</h4>
<img data-src="../assets/fitland_cartoon.svg" width=50%>
$
f(x_1,\, \ldots,\, x_p) = \beta_0 + \sum_{i} x_i \beta_i
+ \sum_{i \le j} x_i x_j \beta_{i,j}
$
$
+ \sum_{i \le j \le k} x_i x_j x_k \beta_{i,j,k}
+ \ldots
$<br>
</section>

<section>
<img data-src="../assets/fitlands_outline.png" width=60%>
</section>

<section>
<h4>State of the art in 2018</h4>
<img data-src="../assets/GlinternetVSxyz.svg" width=80%></br>
<font size="3">
<a href="https://doi.org/10.1371/journal.pone.0254491">
Kieran Elmes et al. Learning epistatic gene interactions from perturbation screens. 2021
</a>
</font>
</section>

<section>
<h4>Problem #1: The hierarchy assumptions</h4>
<img data-src="https://upload.wikimedia.org/wikipedia/commons/4/4f/Manhattan_plot_from_a_GWAS_of_kidney_stone_disease.png"><br>
<font size="3">
<a href="https://doi.org/10.1038/s41467-019-13145-x">
Sarah Howles et al. Genetic variants of calcium and vitamin D metabolism in kidney stone disease. 2019</a>
</font>
</section>

<section>
<h4>Problem #2: Scalability</h4>
<img data-src="../assets/glinternet_xyz_runtime.svg"><br>
<font size="3">
<a href="https://doi.org/10.1371/journal.pone.0254491">
Kieran Elmes et al. Learning epistatic gene interactions from perturbation screens. 2021
</a>
</font>
</section>

<section>
<font size="6">
Hierarchy and scalability are largely solved (until you sequence more)
</font>
<img data-src="../assets/pint_summary.svg"><br>
<font size="3">
<a href="https://doi.org/10.1371/journal.pcbi.1010730">
Kieran Elmes et al. A fast lasso-based method for inferring higher-order interactions
</a>
</font>
</section>

<section>
<h3>Problem #3 (unsolved!):</h3>
<h3>Model assumptions</h3>
$
f(x_1,\, \ldots,\, x_p) = \beta_0 + \sum_{i} x_i \beta_i
+ \sum_{i \le j} x_i x_j \beta_{i,j}
$
$
+ \sum_{i \le j \le k} x_i x_j x_k \beta_{i,j,k}
+ \ldots
$<br>
<font size="3">
<a href="https://doi.org/10.1371/journal.pcbi.1010730">
Kieran Elmes et al. A fast lasso-based method for inferring higher-order interactions
</a>
</font>
</section>

<section>
<h3>SNVformer</h3>
<img data-src="../assets/snvformer_predictions.svg"><br>
<font size="3">
<a href="https://icml-compbio.github.io/2022/papers/WCBICML2022_paper_58.pdf">
Kieran Elmes et al. SNVformer: an attention-based deep neural network for GWAS data
</a>
</font>
</section>

<section>
<h4>"eQTL" simulations (work in progress)</h4>
<img data-src="../assets/eqtl_simulation.svg">
<font size="3">
<ul>
<li >Multi-Layer Perceptron [ AI ]
<li> Logistic Regression [ :) ]
<li> Transformer-encoder [ AI ]
<li> LightGBM (Random Forests) [ AI? ]
<li> Differentiable Logic [ AI ]
<li> Support Vector Machines [ AI? ]
</ul>
</font>
</section>

<section>
<h4>Why don't you just do AI?</h4>
<img data-src="../assets/linformer_model_selection.png" width=60%><br>
What AI?
</section>

<section>
<h4><a href="https://biods.org">bioDS lab @UCNZ</a></h4>
<img data-src="../assets/2023-02-17-bioDS_lab.jpg" width=75%>
</section>

<section>
<h4>Thank you!</h4>
<img data-src="../assets/thanks_rdf_marsden_mbie.svg">
</section>

</div>
</div>


<script src="../talks/revealjs/dist/reveal.js"></script>
<script src="../talks/revealjs/plugin/markdown/markdown.js"></script>
<script src="../talks/revealjs/plugin/highlight/highlight.js"></script>
<script src="../talks/revealjs/plugin/notes/notes.js"></script>
<script src="../talks/revealjs/plugin/math/math.js"></script>
<script>
Reveal.initialize({
controls: true,
progress: true,
history: true,
center: true,
slideNumber: 'h.v',
plugins: [ RevealMarkdown, RevealHighlight, RevealNotes, RevealMath.MathJax2 ]
});
</script>

<script>
(function(i,s,o,g,r,a,m){i['GoogleAnalyticsObject']=r;i[r]=i[r]||function(){
(i[r].q=i[r].q||[]).push(arguments)},i[r].l=1*new Date();a=s.createElement(o),
m=s.getElementsByTagName(o)[0];a.async=1;a.src=g;m.parentNode.insertBefore(a,m)
})(window,document,'script','https://www.google-analytics.com/analytics.js','ga');

ga('create', 'UA-86783531-1', 'auto');
ga('send', 'pageview');
</script>
</body>
</html>

0 comments on commit 969eaad

Please sign in to comment.