Step 1: Data Quality Control and Preprocessing

Step 1a: Cell Ranger Pipeline - Initially, the raw sequencing reads were processed using the Cell Ranger software (10X Genomics), which handles barcode processing, alignment of reads to the reference genome, and quantification of gene expression levels.

Step 1b: Filtering - The initial output includes a matrix of gene counts across individual nuclei. This matrix was then filtered to remove low-quality nuclei and potential doublets, ensuring that the data used for further analysis represented high-quality, single-nucleus profiles.

Step 2: Data Integration and Normalization

Step 2a: Seurat Package - Using the R package Seurat, data from ADPKD and control samples were integrated to correct for batch effects, a common issue when sequencing data are collected across multiple experiments.

Step 2b: Normalization - Within Seurat, data normalization procedures were applied to mitigate the impact of varying sequencing depths across samples and to standardize the expression measurements for comparative analysis.

Step 3: Dimensionality Reduction and Clustering

Step 3a: Principal Component Analysis (PCA) - This step reduces the complexity of the data by transforming the high-dimensional dataset (genes x cells) into a lower-dimensional space while preserving as much variance as possible.

Step 3b: Clustering and UMAP Visualization - Further dimensional reduction was performed using Uniform Manifold Approximation and Projection (UMAP) for visualization. This method helps in visualizing clusters of nuclei that share similar expression profiles, indicative of specific cell types or states.

Step 4: Differential Expression Analysis

Step 4a: Identification of Differentially Expressed Genes - For each cluster identified in the previous step, differential expression analysis was conducted to identify genes that were significantly up- or down-regulated in ADPKD samples compared to controls.

Step 4b: Marker Gene Identification - This analysis also facilitated the identification of marker genes for each cell cluster, aiding in the characterization and annotation of cell types present in the kidney samples.

Step 5: Pathway and Network Analysis

Step 5a: Enrichment Analysis - Using gene set enrichment analysis (GSEA) tools like MSigDB, the researchers identified key signaling pathways and biological processes that were enriched in specific clusters, particularly those altered in ADPKD.

Step 5b: Network Construction - Interaction networks were constructed to identify potential signaling pathways and interactions between identified markers, providing insights into the molecular mechanisms driving ADPKD.

R Code

x <- read_tsv("https://borreliabase.org/~wgqiu/muto1.tsv")
head(x)

## # A tibble: 6 × 2,002
##   id        cell.type   PTPRQ   PTPRO ST6GALNAC3  MAGI2 SERPINE1  SLC8A1 SLC26A7
##   <chr>     <chr>       <dbl>   <dbl>      <dbl>  <dbl>    <dbl>   <dbl>   <dbl>
## 1 PKD_ACAG… PT2       -0.258  -0.338      -0.359 -1.49     1.85  -0.968   -0.405
## 2 PKD_ACCA… DCT       -0.385  -0.523      -0.422  1.63    -0.421  0.500   -0.251
## 3 PKD_ACGC… DCT       -0.210  -0.355      -0.536  0.490   -0.338 -0.828   -0.297
## 4 PKD_ACGC… z12       -0.798  -0.966       2.10   0.827   -0.199  0.0483  -0.308
## 5 PKD_ACGG… PT1        1.56   -0.211      -0.400 -0.479    0.431  1.81    -0.372
## 6 PKD_ACGG… TAL2      -0.0403  0.0271     -0.365 -1.32    -0.330 -0.256   -0.203
## # ℹ 1,993 more variables: PLA2R1 <dbl>, SLC12A3 <dbl>, AC109466.1 <dbl>,
## #   EMCN <dbl>, NTNG1 <dbl>, TIMP3 <dbl>, ZPLD1 <dbl>, LINC01811 <dbl>,
## #   CELF2 <dbl>, PLAT <dbl>, ROBO2 <dbl>, FGF1 <dbl>, FMN2 <dbl>,
## #   SLC14A2 <dbl>, PAPPA2 <dbl>, PLPP1 <dbl>, ADAMTS19 <dbl>, CLNK <dbl>,
## #   LDB2 <dbl>, AC093912.1 <dbl>, ADGRL3 <dbl>, NKAIN2 <dbl>, CLIC5 <dbl>,
## #   NPHS2 <dbl>, ADGRF5 <dbl>, FLT1 <dbl>, ADAMTS6 <dbl>, RGS6 <dbl>,
## #   DCC <dbl>, SNTG1 <dbl>, CNTNAP5 <dbl>, ENPP2 <dbl>, NPHS1 <dbl>, …

R Code

x <- x %>% 
  mutate(cohort = str_remove(id, "_.+")) %>% 
  mutate(sample = str_remove(id, "_.+_") )

Total Samples

The dataset comprises 12 distinct samples, segmented into two main categories: 5 control samples (labelled CONT1 through CONT5) and 7 samples associated with Polycystic Kidney Disease (PKD), designated as PKD1 through PKD7.

Cells Per Sample

Each sample consistently contains 200 cells, facilitating uniform data analysis across different conditions.

Variety of Cell Types

A total of 13 unique cell types have been identified, enhancing the dataset’s utility for studying cellular diversity. The cell types include:

1. CNT_PC: Connecting Tubule and Principal Cells
2. DCT: Distal Convoluted Tubule
3. ENDO: Endothelial Cells
4. FIB: Fibroblasts
5. ICA_ICB: Intercalated Cells Types A and B
6. LEUK: Leukocytes
7. PEC: Parietal Epithelial Cells
8. PODO: Podocytes
9. PT1: Proximal Tubule Type 1
10. PT2: Proximal Tubule Type 2
11. TAL1: Thick Ascending Limb of Henle’s Loop Type 1
12. TAL2: Thick Ascending Limb of Henle’s Loop Type 2
13. z12: A unique cell type

R Code

table(x$cohort)

## 
## Cont  PKD 
## 1000 1400

table(x$sample)

## 
## Cont1 Cont2 Cont3 Cont4 Cont5  PKD1  PKD3  PKD4  PKD5  PKD6  PKD7  PKD8 
##   200   200   200   200   200   200   200   200   200   200   200   200

table(x$cell.type)

## 
##  CNT_PC     DCT    ENDO     FIB ICA-ICB    LEUK     PEC    PODO     PT1     PT2 
##     352     204     104     188     122     123      45      45     343     189 
##    TAL1    TAL2     z12 
##     172     483      30

R Code

MD_Data.long <- x %>%
  pivot_longer(
    cols = 3:2002,        
    names_to = "gene",    
    values_to = "rna.ct")

Column & Row

In a heatmap generated from gene expression data, the columns represent individual genes. Each column corresponds to a specific gene whose expression levels are measured across different samples or conditions. The selection of these genes might be based on variability, relevance to the study, or other criteria depending on the analysis purpose. Each row corresponds to a different biological sample or experimental condition. For example, in a study of disease, rows could represent different patient samples, or in a time-course study, different time points.

What is the color gradient?

The color gradient in a heatmap represents the scale of measurement values, commonly gene expression levels in this context. Colors vary typically from low to high values, where: One end of the color spectrum might represent lower expression levels. The middle of the spectrum (e.g., white) often represents average or median expression levels. The other end of the spectrum represents higher expression levels.

What is the color bar for the rows?

A color bar for the rows, known as row annotation, can be used to provide additional information about each row. This could represent categorical data such as the cell type, treatment group, disease status, or any other pertinent classification of the samples. Each category is typically represented by a different color, and this bar runs parallel to the rows, helping to visually associate samples with their metadata.

What do the cluster diagrams represent?

The cluster diagrams, or dendrograms, displayed on the sides of the heatmap, represent hierarchical clustering results. They are visual representations of the distance or similarity between the elements (samples or genes) based on their expression patterns:

Dendrogram Meanings

The horizontal dendrogram shows clustering of columns, i.e., how genes cluster together based on their expression profiles across all samples. Genes that show similar expression patterns across the samples are clustered together.

The vertical dendrogram shows clustering of rows, i.e., how samples cluster together based on their gene expression across all measured genes. Samples with similar expression profiles are grouped, highlighting potential similarities in biological or experimental conditions.

These dendrograms help in identifying groups of genes or samples that behave similarly under the study conditions, potentially revealing biological processes or subgroup characteristics within the data.

PCA Plot of Kidney Cell Types

Each point represents a single cell’s gene expression profile projected onto the first two principal components (PC1 and PC2), which capture the largest variances within the dataset. PC1 explains the most variance among the samples, followed by PC2, which is orthogonal to PC1. The colors distinguish different cell types, highlighting the gene expression variability and clustering patterns among them. The proximity of points indicates similarity in gene expression profiles; closer points suggest similar profiles, while more distant points indicate dissimilar profiles. Ellipses represent statistical confidence regions for each cell type, aiding in visualizing the grouping and spread of data points within each category.

UMAP Plot of Kidney Cell Types

UMAP Dimension 1 is the first dimension resulting from the UMAP algorithm. It represents the primary axis of variance in the dataset, capturing the largest difference across the data according to the manifold approximation. UMAP Dimension 2 is the second dimension resulting from the UMAP algorithm. It captures the second most significant variance in the dataset, orthogonal to the first dimension.

Each point in the UMAP plot represents an individual observation from the dataset. Each point corresponds to a single cell’s data profile based on its gene expression or other molecular features being analyzed. The location of each point in the plot reflects the cell’s relationship to other cells based on these features. Colors in the UMAP plot are used to indicate categories within the data. In biological datasets: Each color represents a different cell type or classification. This allows for quick visual differentiation between groups of cells and helps identify patterns or clusters within specific cell types.

Distances in a UMAP plot represent a non-linear approximation of similarity between data points. These points that are close together in the UMAP space exhibit similar gene expression profiles or molecular characteristics. They are more alike in terms of the data set’s measured features.Conversely, cells that are far apart have dissimilar profiles, indicating a significant difference in their gene expression or other key features analyzed.

Color Interpretation

Gray (NS): Represents genes that are not statistically significant. These genes do not meet the criteria set for significance in terms of p-value and fold change.

Green: Represents genes with significant log2 fold changes but not significant p-values. These genes show considerable changes in expression levels but are not statistically significant.

Blue: Represents genes that are statistically significant in terms of p-value only. These genes have low p-values but their fold changes are not beyond the set threshold for log2 fold change.

Red: Represents genes that are significant both in terms of p-value and log2 fold change. These are the genes of highest interest because they show both a statistically significant difference in expression and a considerable magnitude of change.

Gene Regulation Interpretation

From the plot, you can identify genes that are significantly up- or down-regulated based on their position. For example, genes such as “ERBB4” and “LINC01811” appear in red, indicating they are significantly regulated. “ERBB4” is shown on the right side of the plot with a positive log2 fold change, indicating significant upregulation. “LINC01811” is also upregulated.

T-Test Results Consistency

To confirm the consistency of these results with statistical analyses like t-tests, we’d need access to the p-values comparing each gene’s expression across cell types. If a gene shows clear differentiation in expression levels between groups visually (e.g., more concentrated in one cell type versus others), it may align with a low p-value from a t-test or other differential expression analysis.Based on visual inspection, “AC019068.1” appears to have the most distinct pattern, with significant expression levels in z12 and PEC cell types while remaining low in most others. Therefore, this gene might be the best biomarker for distinguishing those target cell types.

Valid marker genes appear to be present. For example, “AC019068.1” shows specific high expression in a limited number of cell types. This distinction is crucial in identifying target cell types from a larger mixture, and such differentiation makes these genes potential biomarkers.

Color Interpretation

Blue: Genes considered significant according to the p-value cutoff (pCutoff = 0.01) and shown in the legend as “p-value”.

Gray: Genes that are not statistically significant.

Gene Regulation Interpretation

A gene that is significantly up-regulated (e.g., ATP1B3 in the provided volcano plot) will have high positive mean differences and be colored in blue. A gene that is significantly down-regulated (e.g., DLGAP1 if the data points exist in the negative region) will have a high negative mean difference.

Graph Interpretation

The x-axis indicates RNA count (rna.ct), which is the expression level of the genes across different cell types. For each cell type, the spread along the x-axis shows the distribution of expression levels (or counts) for a specific gene. The y-axis lists various cell types (e.g., z12, TAL2, PT1). Each box plot is organized vertically to show how a particular gene’s expression varies across all these cell types. The individual points represent specific observations (samples) within a given cell type. The points are slightly jittered horizontally to prevent overlap and improve visibility, while boxplot overlays indicate the median and interquartile ranges. The red and teal colors differentiate between cohorts. Each panel represents a specific gene, with its name displayed at the top of the panel. This layout helps visualize the expression patterns of multiple genes side by side across the same cell types, highlighting differential expression.

T-Test Results Consistency

The panels showing violin plots are consistent with a statistical test such as a t-test.A gene showing clear separation in expression levels between cohorts (red vs. teal points) would likely yield significant p-values in statistical analysis. “LINCO1811” shows a distinctive pattern with clear separation between the two cohorts, suggesting significant differential expression.”LINCO1811” appears to be a strong candidate for a biomarker for PKD patients. It shows significant upregulation in the PKD cohort (teal) compared to the control cohort (red), especially within certain cell types (PT1 and PT2).

Experimental Procedure

The use of snRNA-seq particularly caters to the challenges of kidney tissue analysis, allowing for robust gene expression profiling even in less viable post-thaw cells. This technology is instrumental in capturing a snapshot of gene activity within individual nuclei, providing a detailed molecular understanding that surpasses traditional bulk tissue analyses.

Computational Analysis

The computational steps outlined, including data quality control, normalization, and clustering, ensure that the data integrity is maintained and that the biological variability is accurately captured. This meticulous processing is crucial for the reliable identification of cell-specific gene expression patterns and the subsequent analysis of these patterns across different cell types and disease states.

Data Visualization and Clustering

The data integration and normalization procedures set the stage for effective comparison across ADPKD and healthy control samples. Dimensionality reduction (PCA and UMAP) and clustering reveal distinct groups of cells, highlighting the cellular heterogeneity within and between ADPKD and healthy samples. The differential expression analysis identifies genes that are significantly up- or down-regulated in ADPKD samples compared to controls, pinpointing potential biomarkers and therapeutic targets.

Pathway and Network Analysis

Enrichment analyses further elucidate the biological pathways that are disproportionately affected in ADPKD, particularly those related to cell proliferation and fibrosis, supporting the biological hypothesis. Network analyses potentially reveal interactions between key proteins and regulatory elements involved in ADPKD, offering insights into the disease’s molecular drivers.

Scientific and Clinical Implications:

The results provide a granular view of the disease mechanisms at the cellular level, which can significantly enhance the understanding of ADPKD progression and its underlying molecular drivers.

The identification of specific gene expression profiles associated with disease states presents new avenues for the development of diagnostic and prognostic biomarkers. The study highlights potential targets for therapeutic intervention, which could lead to the development of more effective treatments tailored to disrupt key pathological pathways in ADPKD.