fix crossrefs

_quarto.yml

@@ -58,4 +58,11 @@ format:
     grid: 
       sidebar-width: 350px
       margin-width: 350px
+
+  # pdf:
+  #   geometry:
+  #     - bottom=17mm
+  #     - left=15mm
+  #     - right=15mm
+  #   fontsize: 16pt
 

docs/survomics.html

@@ -465,7 +465,7 @@ <h2 class="unnumbered anchored" data-anchor-id="tcga-omics-data">TCGA omics data
 </div>
 <div class="callout-body-container callout-body">
 <ul>
-<li><p><a href="https://bioconductor.org/packages/release/bioc/html/TCGAbiolinks.html">Bioconductor</a> might provide an outdated version of <strong>TCGAbiolinks</strong>. Here, we use the GitHub version TCGAbiolinks_2.29.6. If you encounter some issues when using this tutorial, please check your installed <strong>TCGAbiolinks</strong> version. If necessary, you can re-install the package from its <a href="https://github.com/BioinformaticsFMRP/TCGAbiolinks.git">GitHub repository</a>. Otherwise, download the data from <a href="https://doi.org/10.5281/zenodo.10044741"><img src="https://zenodo.org/badge/DOI/10.5281/zenodo.10044741.svg" class="img-fluid" alt="DOI"></a> and load the <code>dat</code> object with: <code>load("TCGA_data.rda")</code>.</p></li>
+<li><p><a href="https://bioconductor.org/packages/release/bioc/html/TCGAbiolinks.html">Bioconductor</a> might provide an outdated version of <strong>TCGAbiolinks</strong>. Here, we use the GitHub version TCGAbiolinks_2.29.6. If you encounter some issues when using this tutorial, please check your installed <strong>TCGAbiolinks</strong> version. If necessary, you can re-install the package from its <a href="https://github.com/BioinformaticsFMRP/TCGAbiolinks.git">GitHub repository</a>. Otherwise, download the data from <a href="https://doi.org/10.5281/zenodo.10044741">here</a> and load the <code>dat</code> object with: <code>load("TCGA_data.rda")</code>.</p></li>
 <li><p>The package <strong>TCGAbiolinks</strong> cannot retrieve any proteomics or metabolomics data. It is always useful to look at your data first, in particular the data type and dimensions (i.e.&nbsp;numbers of rows and columns for a data frame or matrix).</p></li>
 </ul>
 </div>
@@ -2050,15 +2050,15 @@ <h4 class="unnumbered anchored" data-anchor-id="feature-stability-analysis-1">Fe
 </section>
 </section>
 </section>
-<section id="multi-omics-integrative-modeling" class="level1 unnumbered">
-<h1 class="unnumbered">Multi-omics integrative modeling</h1>
+<section id="multi-omics-integrative-modeling" class="level2 unnumbered">
+<h2 class="unnumbered anchored" data-anchor-id="multi-omics-integrative-modeling">Multi-omics integrative modeling</h2>
 <p>Integration of multi-omics data from various types of omics profiling technologies can improve our understanding of complex disease mechanisms holistically, and hence improve prediction of disease progression and survival <span class="citation" data-cites="Hasin2017 Subramanian2020 Vandereyken2023">(<a href="#ref-Hasin2017" role="doc-biblioref">Hasin, Seldin, and Lusis 2017</a>; <a href="#ref-Subramanian2020" role="doc-biblioref">Subramanian et al. 2020</a>; <a href="#ref-Vandereyken2023" role="doc-biblioref">Vandereyken et al. 2023</a>)</span>. One obvious benefit of multi-omics data is the availability of biological information from multiple, partly redundant levels, i.e., different modalities, which allows for the assessment of associations within and between different omics datasets.</p>
 <p>A naive strategy for multi-omics data analysis is to treat all the omics features equally and independently, which may lead to a worse performance in comparison to methods that take into account the group structure. A recent large-scale benchmarking study for survival prediction using TCGA multi-omics data demonstrated a slightly improved survival prediction performance when taking into account the group structure of the multi-omics data sets <span class="citation" data-cites="Herrmann2021">(<a href="#ref-Herrmann2021" role="doc-biblioref">Herrmann et al. 2021</a>)</span>. However, <span class="citation" data-cites="Herrmann2021">Herrmann et al. (<a href="#ref-Herrmann2021" role="doc-biblioref">2021</a>)</span> did not evaluate the survival model performance in terms of feature selection, i.e., selection of the most prognostic omics variables or features. To capture associations within modalities of omics features, penalized regressions with a <span class="math inline">\(\ell_2\)</span>-norm penalty or Bayesian methods with a group lasso prior can become useful. Focusing on feature selection or parsimonious effects is often beneficial for the purpose of clinical implementation, where one can use the sparse Group-Lasso Cox <span class="citation" data-cites="Simon2013">(<a href="#ref-Simon2013" role="doc-biblioref">N. Simon et al. 2013</a>)</span> and Bayesian Cox with elastic priors <span class="citation" data-cites="Lee2015">(<a href="#ref-Lee2015" role="doc-biblioref">Lee, Chakraborty, and Sun 2015</a>)</span>.</p>
 <p>To capture associations between modalities of omics features, including their overlapping or nested and hierarchical relationships, the use of biological network structures can become useful, as for example in the Bayesian Cox model with spike-and-slab and MRF priors <span class="citation" data-cites="Madjar2021">(<a href="#ref-Madjar2021" role="doc-biblioref">Madjar et al. 2021</a>)</span>. There is an increasing knowledge of biological interconnections across various molecular profiles, and systems biology approaches are being developed that try to capture these deep insights when modeling or predicting disease mechanisms <span class="citation" data-cites="Yan2018 Karimi2022">(<a href="#ref-Yan2018" role="doc-biblioref">Yan et al. 2018</a>; <a href="#ref-Karimi2022" role="doc-biblioref">Karimi et al. 2022</a>)</span>. There is a number of review and benchmarking studies of the methodologies for multi-omics integration, although these reviews do not focus specifically on survival applications; for example, <span class="citation" data-cites="Herrmann2021">Herrmann et al. (<a href="#ref-Herrmann2021" role="doc-biblioref">2021</a>)</span> for penalized regression-based methods, boosting-based methods and random forest-based methods; <span class="citation" data-cites="Agamah2022">Agamah et al. (<a href="#ref-Agamah2022" role="doc-biblioref">2022</a>)</span> for network analyses, <span class="citation" data-cites="Ickstadt2018">Ickstadt, Schäfer, and Zucknick (<a href="#ref-Ickstadt2018" role="doc-biblioref">2018</a>)</span> and <span class="citation" data-cites="Chu2022">Chu et al. (<a href="#ref-Chu2022" role="doc-biblioref">2022</a>)</span> for Bayesian approaches; and <span class="citation" data-cites="Kang2022">Kang, Ko, and Mersha (<a href="#ref-Kang2022" role="doc-biblioref">2022</a>)</span> for deep learning methods.</p>
 <p>However, to achieve a comprehensive and biologically meaningful integration of high-dimensional multi-omics data, there is a need for continued development of computational and statistical approaches that consider both technical and biological intricacies of the data and technologies, respectively <span class="citation" data-cites="Wissel2023">(<a href="#ref-Wissel2023" role="doc-biblioref">Wissel, Rowson, and Boeva 2023</a>)</span>. This is currently a very active research field, and we expect to see many improved multi-omics methods for survival prediction in the future.</p>
 </section>
-<section id="r-session-info" class="level1 unlisted unnumbered">
-<h1 class="unlisted unnumbered">R session info</h1>
+<section id="r-session-info" class="level2 unlisted unnumbered">
+<h2 class="unlisted unnumbered anchored" data-anchor-id="r-session-info">R session info</h2>
 <div class="cell">
 <div class="sourceCode cell-code" id="cb105"><pre class="sourceCode r code-with-copy"><code class="sourceCode r"><span id="cb105-1"><a href="#cb105-1" aria-hidden="true" tabindex="-1"></a><span class="fu">sessionInfo</span>()</span></code><button title="Copy to Clipboard" class="code-copy-button"><i class="bi"></i></button></pre></div>
 </div>
@@ -2161,8 +2161,8 @@ <h1 class="unlisted unnumbered">R session info</h1>
 [181] biomaRt_2.56.1              rlang_1.1.1                 uuid_1.1-1                 
 [184] fansi_1.0.4                 prodlim_2023.03.31          psbcSpeedUp_2.0.4</code></pre>
 </section>
-<section id="references" class="level1">
-<h1>References</h1>
+<section id="references" class="level2">
+<h2 class="anchored" data-anchor-id="references">References</h2>
 
 
 <div id="refs" class="references csl-bib-body hanging-indent" role="list">

survomics.qmd

@@ -214,7 +214,7 @@ RNA_count <- RNA_count[, rownames(clin)]
 Here, we use the GitHub version TCGAbiolinks_2.29.6. 
 If you encounter some issues when using this tutorial, please check your installed **TCGAbiolinks** version. 
 If necessary, you can re-install the package from its [GitHub repository](https://github.com/BioinformaticsFMRP/TCGAbiolinks.git).
-Otherwise, download the data from [![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.10044741.svg)](https://doi.org/10.5281/zenodo.10044741) and load the `dat` object with: `load("TCGA_data.rda")`.
+Otherwise, download the data from [here](https://doi.org/10.5281/zenodo.10044741) and load the `dat` object with: `load("TCGA_data.rda")`.
 
 - The package **TCGAbiolinks** cannot retrieve any proteomics or metabolomics data.
 It is always useful to look at your data first, in particular the data type and dimensions (i.e. numbers of rows and columns for a data frame or matrix).
@@ -2110,7 +2110,7 @@ tibble::tibble(jaccard = jac, nogueira = nog, zucknick = zuck)
 
 From the above values we conclude that the stability of Lasso Cox's feature selection is neither poor nor excellent but somewhere in between.
 
-# Multi-omics integrative modeling {-}
+## Multi-omics integrative modeling {-}
 
 Integration of multi-omics data from various types of omics profiling technologies can improve our understanding of complex disease mechanisms holistically, and hence improve prediction of disease progression and survival [@Hasin2017;@Subramanian2020;@Vandereyken2023]. 
 One obvious benefit of multi-omics data is the availability of biological information from multiple, partly redundant levels, i.e., different modalities, which allows for the assessment of associations within and between different omics datasets. 
@@ -2128,7 +2128,7 @@ There is a number of review and benchmarking studies of the  methodologies for m
 However, to achieve a comprehensive and biologically meaningful integration of high-dimensional multi-omics data, there is a need for continued development of computational and statistical approaches that consider both technical and biological intricacies of the data and technologies, respectively [@Wissel2023]. 
 This is currently a very active research field, and we expect to see many improved multi-omics methods for survival prediction in the future.
 
-# R session info {.unlisted .unnumbered}
+## R session info {.unlisted .unnumbered}
 
 ```{r}
 sessionInfo()
@@ -2238,5 +2238,5 @@ loaded via a namespace (and not attached):
 knitr::write_bib("knitr", "references.bib")
 ```
 
-# References 
+## References