<div id="links"> Slides: https://go.wisc.edu/669t74<br> Lab Site: https://measurement-and-microbes.org </div> <div style="height: 2em;"></div> ### Interactive and Interpretable Omics Analysis My lab writes statistical software that helps biologists get the most out of their data. The essential questions are, * **Experimental Design**: How should we design experiments to accelerate engineering or medical applications? * **Integration**: How should we analyze data gathered from multiple batches or technologies? * **Reproducibility**: How can we be sure our conclusions are trustworthy? We work directly with microbiologists on problems related to HIV, the gut-brain axis, and synthetic communities. --- exclude: true ### Problem Solving We strive to use advanced multiomics workflows while keeping the analysis transparent. To this end, we draw from the data visualization and machine learning interpretability literatures. * **Visualization**: A good interface can shape the way we think for the better, helping us be more critical and creative. * **Interpretability**: We want models that we can trust and control. --- ### Themes: Visualization Stacked bar plots are used in microbiome research to compare composition across groups or over time. .pull-left[ - `\(x\)`-axis position encodes samples - Color encodes taxonomy ] .pull-right[ <img src="data:image/png;base64,#figures/qiime_taxonomic_barplot_nature.png" width="80%"/> Figure from the Qiime2 paper [1]. ] --- ### Themes: Visualization The basic limitation is that they can only show so many colors at a time. This makes it difficult to study rarer taxa. .center[ <img src="data:image/png;base64,#figures/qiime_barplot.png" width=530/> ] Figure from the [Qiime2 documentation](https://view.qiime2.org/visualization/?src=https://docs.qiime2.org/2024.2/data/tutorials/moving-pictures/taxa-bar-plots.qzv). --- ### Themes: Visualization Undergraduate Megan Kuo wrote the phylobar package to create color palettes by painting an accompanying tree structure [2]. .center[ <img src="data:image/png;base64,#figures/phylobar_overview.png" width=600/> ] Figure from the [phylobar documentation](https://mkdiro-o.github.io/phylobar/articles/hfhs.html#htmlwidget-ac96cb3ee4656e2e9ec3) [2]. --- ### Pseudobulk Data This idea works whenever you cared about hierarchically organized compositions, like this COVID-19 immunology data.  --- ### Themes: Visualization In [3], we used the focus-plus-context principle to interactively explore how dimensionality reduction methods like UMAP and t-SNE distort the intrinsic geometry of the data. .center[ <img src="data:image/png;base64,#figures/hydra_perplexity40_distortion.svg" width=500/> ] --- ### Themes: Visualization In [3], we used the focus-plus-context principle to interactively explore how dimensionality reduction methods like UMAP and t-SNE distort the intrinsic geometry of the data. .center[ <img src="data:image/png;base64,#figures/hydra_perplexity40_distortion_hover.svg" width=500/> ] --- ### Themes: Visualization In [3], we used the focus-plus-context principle to interactively explore how dimensionality reduction methods like UMAP and t-SNE distort the intrinsic geometry of the data. .center[ <img src="data:image/png;base64,#figures/hydra_perplexity40_distortion_hover_neighbor.svg" width=500/> ] --- ### Themes: Visualization In [3], we used the focus-plus-context principle to interactively explore how dimensionality reduction methods like UMAP and t-SNE distort the intrinsic geometry of the data. .center[ <img src="data:image/png;base64,#figures/hydra_perplexity230_distortion.svg" width=500/> ] --- ### Themes: Visualization In [3], we used the focus-plus-context principle to interactively explore how dimensionality reduction methods like UMAP and t-SNE distort the intrinsic geometry of the data. .center[ <img src="data:image/png;base64,#figures/hydra_perplexity230_distortion_hover.svg" width=500/> ] --- ### Themes: Visualization In [3], we used the focus-plus-context principle to interactively explore how dimensionality reduction methods like UMAP and t-SNE distort the intrinsic geometry of the data. .center[ <img src="data:image/png;base64,#figures/hydra_perplexity230_distortion_hover_neighbor.svg" width=500/> ] --- ### Themes: Interpretability We want to benchmark omics analysis tasks like batch integration and trajectory inference. VAEs can generate realistic data, but how can we create ground truth? .center[ <img src="data:image/png;base64,#figures/lung_learned_distribution_simulation_fidelity.png" width=500/> ] --- ### Themes: Interpretability PhD student Langtian Ma found a way to intervene on VAE latent states in a way that supports precise control of both ground truth signal and technical effects. .center[ <img src="data:image/png;base64,#figures/batch_effect_dose_response.png" width=750 /> ] --- ### Collaborations with Biologists * HIV Risk Factors [Kwon Lab, Ragon Institute]: The microbiome is closely related to immunity. We are using data to understand HIV risk from multi-omics data. * Mental Health and the Microbiome [Handelsman Lab, UWM]: Specific microbes have been linked to mood disorders. We are part of a multi-institute collaboration to better understand these gut-brain connections. - We are actively recruiting for this project. --- ### Reaching Out * You can learn more at [https://measurement-and-microbes.org](https://go.wisc.edu/pgb8nl). - Interactivity: [2; 4; 3; 5] - Interpretability: [6; 7; 8; 9; 10] * I enjoy working with students with different educational levels and backgrounds. * Email: [ksankaran@wisc.edu](mailto:ksankaran@wisc.edu) --- class: reference ### References [1] E. Bolyen, J. R. Rideout, M. R. Dillon, N. A. Bokulich, C. C. Abnet, G. A. Al-Ghalith, H. Alexander, E. J. Alm, M. Arumugam, F. Asnicar, et al. "Reproducible, interactive, scalable and extensible microbiome data science using QIIME 2". In: _Nature Biotechnology_ 37.8 (Jul. 2019), p. 852–857. ISSN: 1546-1696. DOI: 10.1038/s41587-019-0209-9. <http://dx.doi.org/10.1038/s41587-019-0209-9>. [2] M. Kuo, K. Lê Cao, S. Kodikara, J. Mao, and K. Sankaran. "phylobar: an R package for multiresolution compositional barplots in omics studies". In: _Bioinformatics_ 42.4 (Mar. 2026). Ed. by M. Nikolski. ISSN: 1367-4811. DOI: 10.1093/bioinformatics/btag151. <http://dx.doi.org/10.1093/bioinformatics/btag151>. [3] K. Sankaran, S. Zhang, Chenab, and M. Meilă. "Interactive visualization of metric distortion in nonlinear data embeddings using the <b> <tt>distortions</tt> </b> package". In: _Briefings in Bioinformatics_ 27.2 (Mar. 2026). ISSN: 1477-4054. DOI: 10.1093/bib/bbag136. <http://dx.doi.org/10.1093/bib/bbag136>. [4] H. Jiang, X. Miao, M. W. Thairu, M. Beebe, D. W. Grupe, R. J. Davidson, J. Handelsman, and K. Sankaran. "Multimedia: multimodal mediation analysis of microbiome data". In: _Microbiology Spectrum_ 13.2 (Feb. 2025). Ed. by J. Claesen. ISSN: 2165-0497. DOI: 10.1128/spectrum.01131-24. <http://dx.doi.org/10.1128/spectrum.01131-24>. [5] K. Sankaran, S. Kodikara, J. J. Li, and K. L. Cao. "Semisynthetic simulation for microbiome data analysis". In: _Briefings in Bioinformatics_ 26.1 (Nov. 2024). ISSN: 1477-4054. DOI: 10.1093/bib/bbaf051. <http://dx.doi.org/10.1093/bib/bbaf051>. [6] K. Sankaran. "Data Science Principles for Interpretable and Explainable AI". In: _Journal of Data Science_ (2024), p. 26–52. ISSN: 1683-8602. DOI: 10.6339/24-jds1150. <http://dx.doi.org/10.6339/24-JDS1150>. [7] R. Yang, Y. Li, K. Sankaran, T. A. Mace, P. A. Hart, Q. Ma, X. Wang, and S. Ke. "Prevalence aware feature selection improves biomarker identification in microbiome studies". In: _Bioinformatics_ 42.7 (Jun. 2026). Ed. by J. Cheng. ISSN: 1367-4811. DOI: 10.1093/bioinformatics/btag371. <http://dx.doi.org/10.1093/bioinformatics/btag371>. [8] J. Fukuyama, K. Sankaran, and L. Symul. "Multiscale analysis of count data through topic alignment". In: _Biostatistics_ 24.4 (Jun. 2022), p. 1045–1065. ISSN: 1468-4357. DOI: 10.1093/biostatistics/kxac018. <http://dx.doi.org/10.1093/biostatistics/kxac018>. [9] A. Janik, J. Dodd, G. Ifrim, K. Sankaran, and K. M. Curran. "Interpretability of a deep learning model in the application of cardiac MRI segmentation with an ACDC challenge dataset". In: _Medical Imaging 2021: Image Processing_. Ed. by B. A. Landman and I. Išgum. SPIE, Feb. 2021, p. 111. DOI: 10.1117/12.2582227. <http://dx.doi.org/10.1117/12.2582227>. [10] K. Sankaran. _Remembrances of States Past_. Observable notebook. Interactive explanation of recurrent neural networks. 2019. <https://observablehq.com/@krisrs1128/remembrances-of-states-past>.