Добавил:
kiopkiopkiop18@yandex.ru t.me/Prokururor I Вовсе не секретарь, но почту проверяю Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:

Ординатура / Хирургия / Библиотека им академика М.И. Перельмана / Книга_5529_Библиотеки_им_академика_М_И_Перельмана

.pdf
Скачиваний:
0
Добавлен:
31.08.2026
Размер:
26 Мб
Скачать
184 Bioinformatics of Autoimmune Diseases
SS
between
SS
Total
is the sum of squares between groups, which measures how much group means devi-where
ate from the overall mean and
is the total sum of squares, which measures the total variation
in the data.
This statistic represents the proportion of the total variance in the dependent variable that is explained by the grouping factor (e.g., treatment or condition). It ranges from 0 to 1, with higher values indicating that a greater portion of the variation is explained by group differences.
In differential expression analysis, η2 indicates how much of the variability in a gene’s expression is explained by the group categories, such as treatment type or disease subtype. A higher η2 suggests that the group factor has a strong inuence on gene expression, making this metric useful for pri­oritizing genes that show consistent and meaningful differences across groups. It helps distinguish between genes that show signicant p-values due to minor variance and those whose expression is genuinely shaped by the grouping variable.
Tukey’s Honest Signicant Difference (HSD) test is a post hoc multiple comparison proce­dure used after ANOVA to determine which specic group means differ from one another. While ANOVA tells us whether there is a statistically signicant difference somewhere among the group means, it does not indicate which groups are different. Tukey’s HSD addresses this by performing pairwise comparisons between all group combinations while controlling for the family-wise error rate. It adjusts the p-values to account for multiple testing and provides condence intervals for each comparison. In the context of gene expression, Tukey’s HSD allows researchers to identify precisely which groups show upregulation or downregulation relative to others, making it especially valuable when analyzing more than two biological conditions or subgroups. It enhances interpretability and helps guide further biological validation or experimental focus.
The summarize _ anova _ results function “summarize _ anova.py” is designed to perform comprehensive differential expression analysis for RNA-Seq data using one-way ANOVA across multiple groups dened by a categorical variable, such as treatment condition, ethnicity, or disease subtype. It begins by preprocessing the input data, ensuring that sample identiers in the expression matrix match those in the metadata design table. For each gene, it collects expres­sion values across the groups, lters out any group with insufcient sample size, and proceeds with ANOVA to determine whether there are signicant differences in expression across the valid groups.
Beyond basic statistical testing, the function computes key effect size metrics that provide greater insight into the magnitude and relevance of observed differences. Eta-squared (η2) is calculated to quantify the proportion of variance in gene expression explained by the grouping variable, offer­ing a sense of how strongly the group classication impacts expression. In addition, when a control group is specied, the function computes log2 fold changes and Cohen’s d for each non-control group in comparison to the control. These values help to interpret not just whether differences exist, but how large and biologically meaningful those differences are. Cohen’s d, in particular, provides a standardized measure of the mean difference relative to variability, making it easier to compare across genes.
If no control group is specied, the function defaults to comparing the top two groups with the greatest mean expression difference for each gene. Alternatively, the function can perform post hoc analysis using Tukey’s Honest Signicant Difference (HSD) test, which performs all pairwise comparisons while adjusting for multiple testing. This option is useful when no control group is dened or when the user wants to explore the full range of pairwise differences across all group combinations.
The output of summarize _ anova _ results is a tidy DataFrame where each row cor- responds to a gene and a group comparison. It includes the gene name, the compared group and baseline, the log2 fold change, Cohen’s d, eta-squared, the A NOVA p-value, and an adjusted p-value that accounts for multiple testing using the Benjamini-Hochberg method. If Tukey post hoc analysis is enabled, the output includes all pairwise comparisons with their respective condence inter­vals and signicance results. This detailed output structure makes it well-suited for downstream
RNA Sequencing 185
TABLE 5.4 Differential Gene Expression Results for Ethnic Groups Relative to “White”
Gene Group Baseline Log2FC Cohen_d Eta_squared p-Value Adj-p
Gene1 Asian White 0.58 0.83 0.17 0.004 0.021 Gene1 Black White 0.12 0.23 0.17 0.004 0.021 Gene2 Asian White 1.04 1.11 0.24 0.001 0.01 Gene2 Black White 0.72 0.78 0.24 0.001 0.01
interpretation, visualization, or integration into gene prioritization workows in studies involving complex biological groupings.
The run _ summarize _ anova.py program is a complete pipeline for performing differen- tial gene expression analysis using one-way ANOVA on normalized RNA-Seq data. It is designed to be modular, exible, and suitable for analyzing experiments with more than two groups. The program begins by reading a normalized count matrix in which rows represent genes and columns represent samples. This matrix is typically generated through preprocessing steps such as align­ment, quantication, and normalization (e.g., using CPM). Alongside the expression matrix, the program reads a study design le in CSV format that contains metadata for each sample, including a unique identier (runID) and several categorical or numeric descriptors such as Ethnic, anti-CCP, or pathotype.
Once the input les are loaded, the program calls the summarize _ anova _ results func- tion, which performs ANOVA-based statistical testing and calculates additional metrics such as log2 fold change, Cohen’s d, and eta-squared (η2). The user can specify which column from the design le should be used as the grouping variable. For example, setting group _ column = 'Ethnic' instructs the program to test whether gene expression varies signicantly across differ­ent ethnic groups. If a control group is dened, such as “White”, the program calculates the log2 fold change and effect size for all other groups relative to that baseline. Alternatively, if no control group is provided and post hoc testing is enabled, the function performs pairwise comparisons using Tukey’s HSD to identify specic group differences with adjusted p-values.
The output of the program is a CSV le that can be used for further analysis or reporting. In a typical use case where a control group is specied, the output table might look like Table 5.4.
In this example, each gene is tested against multiple ethnic groups with respect to the control group. The table includes the effect size and the proportion of expression variance explained by ethnicity, providing insight into both statistical and biological relevance.
In another use case, where post hoc testing is enabled and no control group is specied, the out­put table changes format to reect all pairwise comparisons like Table 5.5.
This output lists the estimated mean difference between groups for each gene, along with con- dence intervals, adjusted p-values, and a ag indicating whether the difference is statistically signicant (reject = True). Such output is particularly useful for exploring all possible group rela­tionships when no natural control group exists.
TABLE 5.5 Tukey’s HSD Pairwise Comparisons of Gene Expression across Ethnic Groups
Gene Group1 Group2 Meandiff p-Adj Lower Upper Reject
Gene1 Asian Black 0.45 0.03 0.02 0.88 TRUE Gene1 Asian White 0.38 0.07 Gene1 Black White
0.07
0.89
0.01
0.47
0.77 FALSE
0.33 FALSE
186 Bioinformatics of Autoimmune Diseases
Overall, the run _ summarize _ anova.py program offers a structured and extensible framework for detecting and interpreting differential gene expression in studies involving complex groupings, and its outputs are suitable for both technical validation and biological discovery.
5.5.7.3 Two-Way ANOVA
Two-way ANOVA is a powerful statistical method used in RNA-Seq data analysis to evaluate the inuence of two independent categorical variables on gene expression, as well as their interaction. Unlike one-way ANOVA, which considers only a single factor, two-way ANOVA allows researchers to explore not only the main effects of each factor individually but also whether there is a signicant interaction effect between them. This makes it particularly useful in complex biological studies where multiple variables may jointly inuence the transcriptional landscape.
In the context of RNA-Seq, two-way ANOVA is implemented by modeling the normalized expression values of each gene across samples, using a linear model that includes both factors and their interaction term. For example, in a study design involving autoimmune disease patients, researchers might be interested in understanding how gene expression is affected by both the patient’s ethnic background and their anti-CCP antibody status. By applying two-way ANOVA, each gene is tested for changes in expression attributable to ethnicity, to anti-CCP status, and to the interaction between these two factors. The analysis typically involves tting a model of the form expression ~ factor1 + factor2 + factor1:factor2 and evaluating the signicance of each term using F-tests.
The use of two-way ANOVA is especially valuable in autoimmune disease research because these conditions often involve multiple contributing factors, both genetic and environmental, that may interact in complex ways. For instance, autoimmune responses may manifest differently across ethnic groups or may vary in severity depending on specic biomarkers like anti-CCP or rheuma­toid factor status. By using two-way ANOVA, researchers can disentangle these effects and identify genes whose expression is modulated not only by individual factors but also by the combination of factors. This can lead to more nuanced insights into disease mechanisms, potential diagnostic mark­ers, and therapeutic targets that are specic to patient subgroups dened by combinations of clinical or demographic variables.
The two-way ANOVA program “rna _ pipeline _ 2wAnova.py” developed for RNA-Seq analysis is a Python-based pipeline designed to identify genes whose expression levels are sig­nicantly inuenced by two categorical variables and their interaction. This is especially useful in biological studies where gene expression is modulated by multiple conditions or patient-specic features. The pipeline begins by either reading in real RNA-Seq data or generating synthetic data for simulation purposes. It ensures that the expression matrix is properly normalized, using CPM, and that the sample metadata contains relevant groupings, such as ethnic background and clinical status like anti-CCP positivity. The design and expression data are then aligned to ensure consis­tency between sample identiers.
At the heart of the analysis lies the two _ way _ anova function, which performs gene-wise sta- tistical testing using a linear model. For each gene, it ts an ordinary least squares model incorporating both main effects (e.g., ethnic group and anti-CCP status) and their interaction term. This allows the pipeline to quantify whether a gene’s expression is affected by one factor independently, by the other, or by a combined interaction of the two. The function collects p-values for each effect and then applies the Benjamini-Hochberg procedure to adjust for multiple testing, yielding adjusted p-values for the main effects and interaction term. These results are collated into a DataFrame and sorted based on the signicance of the interaction effect, which is often of greatest interest in complex diseases.
A sample output from the two-way ANOVA is shown in Table 5.6. Each row represents a gene and includes p-values and adjusted p-values for the ethnic group, anti-CCP status, and their interaction.
This table highlights how specic genes respond to the studied variables. For instance, Gene42 shows signicant effects for both ethnic group and interaction, suggesting its expression is inu­enced not only by group membership but also by how ethnicity and anti-CCP status interact. Such
187 RNA Sequencing
TABLE 5.6 Two-Way ANOVA Results for Gene Expression across Ethnic Groups and Anti-CCP Status
Gene Ethnic_p Anti-CCP_p Interaction_p Ethnic_adj-p Anti-CCP_adj-p Interaction_adj-p
Gene42 0.0012 0.045 0.0035 0.009 0.087 0.018 Gene119 0.032 0.0021 0.089 0.045 0.017 0.135 Gene792 0.0008 0.054 0.0012 0.007 0.091 0.01
ndings are particularly valuable in autoimmune disease research, where genetic background and immune markers often work together to shape disease progression and treatment response.
To enhance interpretability, the pipeline also includes visualizations such as volcano plots, heat­maps, and expression boxplots. The heatmap function, in particular, has been adapted to display group annotations based on the combination of both categorical variables, with color-coded legends that explain group identities. This visualization helps researchers quickly observe clustering patterns and validate whether signicant genes show biologically meaningful differences across groups.
5.5.8 VISUALIZING DIFFERENTIAL GENE EXPRESSION IN RNA-SEQ DATA
Visualizing differential gene expression in RNA-Seq data is a crucial step in understanding how gene activity varies under different biological conditions. Once raw sequencing reads are processed and aligned to a reference genome, statistical methods are used to identify genes that show sig­nicant changes in expression levels between groups, such as diseased versus healthy tissues or treated versus untreated samples. However, these numerical results alone can be overwhelming and difcult to interpret. Visualization helps transform complex datasets into intuitive and insightful representations, allowing researchers to detect patterns, assess data quality, and draw biologically meaningful conclusions. From heatmaps that cluster genes and samples based on expression proles to volcano plots that highlight the magnitude and signicance of gene expression changes, each visualization technique offers a different lens through which to explore the data. Effective visual­izations not only enhance the interpretability of results but also serve as a powerful communication tool in both scientic publications and presentations.
In the following, we discuss several visualization plots that are commonly used for illustrating differential expression in RNA-Seq data.
5.5.8.1 PCA Plot
The PCA (principal component analysis) plot in RNA-Seq analysis is a dimensionality reduction technique used to visualize patterns in high-dimensional gene expression data. It transforms the original gene expression matrix, where each sample is represented by thousands of gene expression values, into a set of uncorrelated variables called principal components. These components are cal­culated based on the directions of maximum variance in the data, with the rst principal component capturing the greatest variation, followed by the second, and so on. To perform PCA, the expression matrix is typically normalized and standardized to ensure comparability across samples. The result is a scatter plot where each point represents a sample, positioned according to its values in the rst two (or more) principal components. The PCA plot helps researchers assess sample clustering, iden­tify outliers, detect batch effects, and understand the inuence of experimental conditions. Samples that cluster closely together are considered to have similar gene expression proles, suggesting simi­lar biological states or treatments, while separation along the principal components may indicate signicant variation due to disease, treatment, or other experimental factors.
The PCA plot program “plot _ pc a.py” is designed to visualize RNA-Seq gene expression data by reducing its complexity and highlighting underlying patterns among samples. It begins by taking a gene expression matrix, where rows represent genes and columns represent samples, along
188 Bioinformatics of Autoimmune Diseases
with a metadata le containing sample-specic information such as treatment group or experimen­tal condition. To prepare the data for PCA, the expression matrix is transposed so that each sample becomes a row, which aligns with the expectations of most machine learning algorithms. The data is then standardized using z-score normalization to ensure that each gene contributes equally to the analysis, regardless of its absolute expression level.
After standardization, the program applies principal component analysis to identify directions of maximum variance in the dataset. It reduces the high-dimensional space to two principal compo­nents, which are used to plot each sample as a point in a two-dimensional scatter plot. Each point is colored based on a user-dened metadata grouping variable, such as condition or treatment, making it easy to observe whether samples from the same group cluster together. Additionally, the program labels each point with the corresponding sample name, using a smaller font to avoid visual clutter while still providing clarity.
The expected output of the program is a high-resolution PNG le containing the PCA plot. This plot includes clear axes labeled with the percentage of variance explained by the rst two principal components, a legend indicating the groupings used for coloring, and a descriptive title. The visual rep­resentation helps researchers quickly identify clusters of similar samples, detect outliers, and evaluate whether biological replicates group as expected. This kind of visualization is crucial in RNA-Seq stud­ies for quality control and to guide further analysis, such as differential expression or batch correction.
In the PCA plot (Figure 5.2), the spatial distribution of points reects the similarity of gene expres­sion proles across samples. Samples that cluster closely together share similar overall expression
FIGURE 5.2 PCA plot of RNA-Seq gene expression data of samples grouped by a nt i - CC P.
189 RNA Sequencing
patterns, while samples that are further apart differ signicantly in their transcriptomic proles. The clear separation between anti-CCP-positive and anti-CCP-negative samples suggests that the anti-CCP condition contributes signicantly to the observed variance in gene expression. This sup­ports the hypothesis that anti-CCP status inuences transcriptional activity in RAand validates the effectiveness of PCA as a tool for identifying biologically meaningful groupings in RNA-Seq data.
5.5.8.2 Volcano Plot
A volcano plot is a type of scatter plot that is commonly used in genomics and transcriptomics to visualize differential expression data. It displays the relationship between statistical signicance and the magnitude of change for each feature, such as genes or transcripts, across different experi­mental conditions. The x-axis of a volcano plot represents the log2 fold change (log2FC), which quanties the magnitude of difference in expression between two groups. The y-axis represents the negative logarithm (base 10) of the p-value or adjusted p-value, typically derived from a statistical test like a t-test or ANOVA. This transformation ensures that features with the most statistically signicant differences appear toward the top of the plot.
To calculate the values for a volcano plot, gene expression data is rst normalized and then subjected to a differential expression analysis using statistical models. Each gene is tested to deter­mine whether its expression differs signicantly between experimental groups. The fold change is calculated as the ratio of mean expression levels between the groups, and the log2 transformation is applied to center the distribution around zero (upregulated genes fall on the right and down­regulated genes on the left). The p-values from the statistical tests are transformed using log10 to emphasize small values, which indicate higher signicance.
Plotting the volcano plot involves creating a scatter plot where each point corresponds to a gene. The x-coordinate of each point is the log2 fold change, and the y-coordinate is the log10 of the p-value. Genes that are both highly differentially expressed and statistically signicant appear as outliers in the upper left and upper right corners of the plot. Researchers often add threshold lines to indicate cutoffs for fold change and signicance (e.g., log2FC greater than 1 or less than 1 and p-value less than 0.05) highlighting genes that pass these criteria in different colors. The resulting volcano plot provides an intuitive visual summary of the data, making it easier to identify genes of interest for further analysis.
The volcano plot program “plot _ volc ano.p y” provides a clear and intuitive visual sum- mary of differential gene expression results from RNA-Seq experiments. In Figure 5.3, gene names have been anonymized (e.g., gene1, gene2) to indicate that the results are illustrative and not intended to represent biologically validated ndings. The core of the volcano plot lies in its ability to show both the magnitude and statistical signicance of gene expression changes. Each gene is plotted with its log2 fold change on the x-axis and the negative log10 of the adjusted p-value on the y-axis. This layout allows researchers to easily identify genes that are both statistically signicant and biologically relevant.
Genes are classied into categories such as “Upregulated”, “Downregulated”, or “Not signi­cant” based on predetermined thresholds for fold change and p-value. These categories are then visually represented using distinct colors (red for upregulated, blue for downregulated, and grey for non-signicant genes) making it easy to discern expression patterns at a glance. Additional lines are drawn to indicate the thresholds used for signicance, which helps to contextualize the position of each gene on the plot.
To enhance interpretability, the script also identies and labels the top ten most signicant genes based on adjusted p-values. These genes are annotated directly on the plot, aiding in the quick identication of key candidates for further analysis or validation. The plot is styled for high-quality output with increased font sizes for labels, legends, and axis titles, making it suitable for both publi­cations and presentations. The nal image is saved in PNG format and is also viewable interactively. This kind of visualization is particularly useful in understanding complex transcriptomic changes in diseases like autoimmune diseases, where many genes may be subtly or strongly affected.
190 Bioinformatics of Autoimmune Diseases
5.5.8.3 MA Plot
The MA plot (minus-average plot) is a powerful visualization used in RNA-Seq differential expres­sion analysis to examine the relationship between the magnitude of gene expression (A) and the change in expression between experimental conditions (M). In the context of RNA-Seq, the A value represents the average expression of a gene across all samples, often calculated as the log2-trans­formed mean of CPM. The M value corresponds to the log2 fold change (log2FC) between two groups, such as disease versus control. Each gene is plotted as a point with its A value on the x-axis and its M value on the y-axis. This allows researchers to quickly assess whether genes with high or low expression levels are differentially expressed between conditions.
To generate the MA plot, normalized expression values are rst obtained for all genes. The aver­age expression for each gene is calculated, and a small constant is added to avoid issues with log transformation. The log2 fold change is derived from statistical testing, typically using t-tests or generalized linear models. Genes are then categorized based on statistical signicance thresholds, such as an adjusted p-value (FDR) cutoff and a log2FC threshold. On the plot, genes that meet these criteria for upregulation or downregulation are highlighted in distinct colors, while non-signicant genes are typically shown in grey. Horizontal lines are often added to indicate fold change thresh­olds and the y = 0 line represents no change between conditions.
MA plots are particularly useful because they highlight systematic patterns and biases in RNA­Seq data. For example, a skewed MA plot may suggest normalization issues or batch effects, while a balanced distribution of signicant genes around the y = 0 line indicates consistent differen­tial expression. Importantly, the MA plot emphasizes how differential expression varies across the dynamic range of expression levels, something that volcano plots, which focus on p-values, may obscure.
In the context of autoimmune diseases, MA plots can be especially benecial for visualizing gene expression differences between patients and healthy controls. Many autoimmune conditions,
191 RNA Sequencing
such as rheumatoid arthritis or lupus, involve complex immune responses that alter gene activ­ity. The MA plot helps identify which genes are upregulated (e.g., cytokines, chemokines, and inammatory regulators) or downregulated in affected individuals. This visual distinction enables researchers to focus on genes with both biologically meaningful changes and robust expression levels. Ultimately, MA plots contribute to a deeper understanding of disease mechanisms, facilitate biomarker discovery, and guide the identication of potential therapeutic targets by clearly present­ing the transcriptional landscape associated with autoimmunity.
The Python code “plot _ MA.p y” for the MA plot is designed to visualize differential gene expression from RNA-Seq data. It uses log2 fold change as the vertical axis (M) and log2-transformed average gene expression as the horizontal axis (A). The input consists of a le containing differ­ential expression statistics, including adjusted p-values and log2 fold changes, and a normalized count matrix. The code rst ensures that gene expression values are uniquely indexed by collapsing duplicate gene symbols via averaging. It then computes the average expression for each gene across all samples and applies a log2 transformation with a small offset to avoid innite values.
Each gene is classied as upregulated, downregulated, or not signicant based on thresholds set for both adjusted p-value and fold change. By default, a gene is considered signicantly upregulated if its log2 fold change exceeds 1 and its adjusted p-value is less than 0.05, and similarly considered downregulated if the log2 fold change is less than 1 with a signicant p-value. If neither condition is met, the gene is labeled not signicant. These categories are visualized using different colors: red for upregulated, blue for downregulated, and grey for non-signicant genes. The code also selects the top most signicant genes based on their adjusted p-values and adds their gene symbols as text labels on the plot to highlight them. Horizontal dashed lines are added to mark the fold change thresholds, and a solid line is drawn at zero to indicate no change.
When interpreting the resulting MA plot (Figure 5.4), genes that lie far above or below the center line with red or blue coloring indicate strong differential expression with statistical support. In the
FIGURE 5.4 MA plot showing differential gene expression in rheumatoid arthritis.
192 Bioinformatics of Autoimmune Diseases
gure, gene names have been anonymized (e.g., gene1, gene2) to indicate that the results are illustra­tive and not intended to represent biologically validated ndings. These genes are typically of most biological interest. However, a common reason why many genes may appear grey, even those that visually cross the threshold lines, is that their adjusted p-values do not meet the specied signi­cance threshold. This is because the code enforces a dual condition for signicance: the gene must exceed both the fold change cutoff and the p-value cutoff to be considered up- or downregulated.
if row['adj-p'] < pval_thresh:
if row['log2FC'] > fc_thresh:
return 'Upregulated'
elif row['log2FC'] < -fc_thresh:
return 'Downregulated'
Therefore, the color coding provides not just a visual sense of expression magnitude but also integrates statistical reliability, making the plot an effective tool for prioritizing candidate genes in further rheumatoid arthritis research.
5.5.8.4 Heatmap
The heatmap for RNA-Seq data provides a visual summary of the expression patterns of selected genes across multiple samples, allowing for the identication of distinct expression proles and potential sample clustering. In this context, the heatmap displays the normalized expression levels of the top differentially expressed genes, standardized across samples to highlight relative differ­ences. Each row corresponds to a gene, and each column represents a sample, with color gradients indicating the level of gene expression. Clustering algorithms are applied to both genes and samples to group those with similar expression proles, making it easier to observe patterns that may be associated with biological conditions such as disease status or treatment response. This visual rep­resentation is especially useful in identifying gene expression signatures that differentiate between experimental groups.
The plot _ heatm ap.py program generates a clustered heatmap to visualize the expression patterns of a selected set of top differentially expressed genes from RNA-Seq data. It begins by sub­setting the normalized expression matrix to include only the specied genes and then standardizes the expression values across samples to emphasize relative differences. The standardized data is transposed so that samples become rows and genes become columns, which is a format compatible with clustering. A hierarchical clustered heatmap is then created using seaborn’s clustermap, apply­ing a diverging colormap to distinguish high and low expression levels. The function also ensures that the output directory exists and saves the plot as a high-resolution PNG le. This visualization helps in exploring how genes and samples group together based on expression patterns, which can be indicative of underlying biological conditions.
This heatmap (Figure 5.5) displays the standardized expression levels of the top ten differentially expressed genes across 20 samples, with 10 samples from anti-CCP positive individuals and 10 from anti-CCP negative individuals. Rows represent samples, and columns represent genes, with color intensity indicating the relative expression level of each gene. Clustering has been applied to both genes and samples, revealing distinct expression patterns that differentiate between the two groups, highlighting potential biomarkers or regulatory pathways associated with rheumatoid arthritis.
5.5.8.5 Gene Expression Plot
The plot _ gene _ expression.py program is designed to visualize the expression levels of a specic gene across different sample groups in an RNA-Seq dataset. It takes as input a normalized count matrix (typically containing log2-transformed CPM values) and a metadata le that contains information about each sample, such as group labels for experimental conditions. The primary pur­pose of the program is to generate a boxplot overlaid with individual data points, allowing users to clearly observe the distribution and variability of gene expression within and between dened groups.
193 RNA Sequencing
FIGURE 5.5 Heatmap of the top ten differentially expressed genes in RA RNA-Seq data.
The program begins by loading the normalized counts and metadata and then extracts expression values for a gene specied by the user. It maps each sample to its corresponding group using the metadata le, enabling group-wise comparison. The resulting plot includes a boxplot that summa­rizes the central tendency and spread of expression for each group, while a strip plot of individual data points is layered on top to show the underlying distribution. This combination offers both sta­tistical and granular views of the gene’s behavior across conditions.
The output of the program (Figure 5.6) is a PNG image of the expression plot, saved with a lename that includes the gene’s name. The plot is annotated with axis labels and a title for clear interpretation, and the style is optimized for readability with appropriate font sizes and spacing. By examining this plot, researchers can assess whether a gene is differentially expressed between groups, whether the variation is consistent across samples, and whether potential outliers exist. This kind of visualization is particularly useful in biomedical research, including autoimmune diseases, where the differential expression of certain genes may indicate immune activation, inammation, or therapeutic response. Such expression plots serve as a vital bridge between statistical analysis and biological interpretation.
5.5.9 FUNCTIONAL ENRICHMENT
Functional enrichment analysis is a powerful approach used to interpret large lists of genes, typi­cally derived from differential expression studies, by identifying biological themes, pathways, or