<b>The loss of insulin-positive cell clusters precedes the decrease of islet frequency and beta cell area in type 1 diabetes</b> by Denise M. Drotar (21679539)

“…Insulin-positive (INS+) single cells (≤10µm), cell clusters (>10 to <35µm), small- and medium-sized islets (35-100µm and 100-200µm) were significantly lost at type 1 diabetes onset, while large INS+ islets (>200µm) were preserved. …”

The three schizophrenia patient clusters visualized using functional connectivity matrices. To create the visualizations, we take all the windows belonging to a cluster and average... by Eloy Geenjaar (21533195)

Subjects:

Construction of the WGCNA Co-expression Network. by Jiyong Zhang (2498740)

Geometric manifold comparison visualization by Eloy Geenjaar (21533195)

Subjects:

Hyperparameter ranges by Eloy Geenjaar (21533195)

Subjects:

Convolutional vs RNN context encoder by Eloy Geenjaar (21533195)

Subjects:

Each subfigure shows a different local size (LS) and context size (CS) configuration, where the reliability of the model across different initializations is measured in R-squared. by Eloy Geenjaar (21533195)

Subjects:

The influence of training-time noise on classification accuracy and reliability by Eloy Geenjaar (21533195)

Subjects:

Data sample demographics. Note that AP refers to anti-psychotic medication, and AD refers to anti-depressive medication. Thus, AP and AD in this table refer to the percentage of pa... by Eloy Geenjaar (21533195)

Subjects:

An abstract depiction of our model, for the independent version of the model, the context representation is not concatenated inside the local encoder. by Eloy Geenjaar (21533195)

Subjects:

These subfigures show a visualization of the context embeddings for the DSVAE model with LS=2=4,CS=2. The first subfigure from the left shows three patient clusters, the second sub... by Eloy Geenjaar (21533195)

Subjects:

The window classification accuracy for two of our proposed models (DSVAE, IDSVAE), windowed FNC (wFNC), and two baseline methods: context-only (CVAE), and local-only (LVAE). by Eloy Geenjaar (21533195)

Subjects:

Summary of non-randomized studies. by Solima Sabeel (9545738)

Loss of Mod causes cell cycle defects. by Amalia S. Parra (4173004)

Mod is required to maintain nucleolar architecture. by Amalia S. Parra (4173004)

Unsupervised clustering of TCGA-KIRC subjects based on glucocorticoid signaling-related genes. by Yu Zhang (12946)

Connectivity between the left nucleus accumbens and two clusters. by Karel Joineau (21799908)

“…<p>A) Representation of the clusters on a 3D brain. Cluster +02 + 16–10 is formed of 255 voxels with 170 voxels covering 15% of the subcallosal cortex, 9 voxels covering 11% of the right nucleus accumbens and 3 voxels covering a non-labelled area. …”

Droplet-based microfluidics can be used to model clustered bacterial growth. by Justin Greene (306625)

Changes in connectivity of the mPFC among responders and non-responders. by Karel Joineau (21799908)

GWAS analysis using Cluster 3 data (shown in S1 Fig). by Ke Li (106849)