RNA-Seq Data Analysis Workflow: From Raw Reads to Insights

RNA sequencing (RNA-seq) has become a standard method for measuring gene expression across entire transcriptomes, but the path from raw sequencing files to biological interpretation requires a structured series of computational steps. This article outlines the complete RNA-seq data analysis workflow, from quality control of raw reads through alignment, quantification, normalization, differential expression testing, and functional interpretation. The focus is on the decision points researchers face at each stage and the criteria for selecting appropriate tools. No single analysis pipeline suits every experimental question, and the choices made at each step depend on the biological question, the organism, the sequencing platform, and the available computational resources.

Scope and Reader Context

This workflow guidance is intended for students, researchers, analysts, and life-science professionals who need to understand the overall structure of RNA-seq data analysis. The content assumes familiarity with basic molecular biology and sequencing concepts but does not require advanced programming expertise. The practical outcome is a clear mental model of the pipeline stages, the key tools available at each stage, and the quality checks that protect against spurious results. Researchers planning their first RNA-seq analysis will find the stage-by-stage structure useful for building their own pipelines, while experienced analysts can use the decision criteria to evaluate alternative approaches.

The workflow described here covers bulk RNA-seq analysis primarily, with attention to single-cell RNA-seq where the analytical steps diverge. The pipeline structure follows the three main stages described in published protocols: preprocessing and data preparation, upstream processing including alignment and quantification, and high-level analyses including differential expression and functional interpretation. Each stage has specific quality control requirements, and skipping these checks can lead to unreliable conclusions.

Experimental Design Considerations Before Analysis Begins

The success of RNA-seq data analysis depends heavily on decisions made before sequencing begins. The experimental design determines what questions can be answered and what statistical power the analysis will have. RNA-seq has a wide variety of applications, and no single analysis pipeline can be used in all cases. The major steps in RNA-seq data analysis include experimental design, quality control, read alignment, quantification of gene and transcript levels, visualization, differential gene expression, alternative splicing, functional analysis, gene fusion detection, and eQTL mapping.

Biological Replication Requirements

The number of biological replicates is the single most important factor in detecting differentially expressed genes reliably. Technical replicates, which involve sequencing the same library multiple times, do not substitute for biological replicates because they do not capture the natural variation between individual organisms or cell populations. The statistical methods used in differential expression analysis estimate biological variability from the replicate samples, and too few replicates produce unstable variance estimates.

For most experimental designs, three biological replicates per condition represent the practical minimum, while five or more replicates provide substantially better power for detecting small expression changes. The relationship between replicate number and statistical power depends on the magnitude of the expression differences, the variability within conditions, and the sequencing depth. Researchers should consider the expected effect sizes in their system when deciding on replicate numbers.

Sequencing Depth and Library Preparation

Sequencing depth, measured as the number of reads per sample, determines how sensitively low-abundance transcripts can be detected. Gene-level analysis requires fewer reads than transcript-level analysis, and detecting splice variants or novel isoforms requires substantially deeper sequencing. The choice between single-end and paired-end reads affects the ability to resolve splice junctions and isoforms, with paired-end reads providing better alignment across exon boundaries.

Library preparation choices, including whether to use poly-A enrichment or ribosomal RNA depletion, influence which RNA species are captured. Poly-A enrichment selects for messenger RNA and is appropriate for most gene expression studies, while rRNA depletion captures a broader range of RNA species including noncoding RNAs. These choices affect the interpretation of the resulting data and should be documented carefully.

Data Management and Reproducibility Planning

Before analysis begins, researchers should establish a data management plan that addresses storage, file naming conventions, and version control for analysis scripts. The FAIR Guiding Principles describe the expectations for making data findable, accessible, interoperable, and reusable. Applying these principles to RNA-seq data means documenting the analysis steps, recording software versions, and depositing both raw data and processed results in appropriate repositories.

Public repositories such as those maintained by the National Center for Biotechnology Information provide infrastructure for depositing and accessing sequencing data. Researchers generating new data should plan for data deposition at the outset of their projects. For studies involving human subjects, the NIH Genomic Data Sharing Policy describes the expectations for data sharing and the protections required for participant privacy. Researchers should review these requirements before generating data to ensure their consent processes and data handling procedures comply.

At a Glance: RNA-Seq Analysis Pipeline Stages

The following table summarizes the major stages of the RNA-seq analysis workflow, the primary inputs and outputs at each stage, and the key decisions researchers must make.

Quality Control of Raw Sequencing Reads

The first stage of any RNA-seq analysis pipeline involves assessing the quality of the raw sequencing data. This step ensures that the data used for downstream analysis is reliable and suitable for subsequent processing. Quality control begins with examining the raw FASTQ files, which contain the sequencing reads along with quality scores for each base.

Base Quality Assessment

The quality scores encoded in FASTQ files represent the probability that each base call is correct. Low-quality bases at the ends of reads are common and can interfere with alignment. Quality assessment tools generate per-base quality plots, per-sequence quality scores, GC content distributions, and adapter contamination reports. These metrics reveal whether the sequencing run performed well and whether any systematic problems affect the data.

The interpretation of quality metrics requires attention to the sequencing platform and library preparation method. Some variation in quality across read positions is expected, but sudden drops in quality or widespread low-quality regions indicate problems that may require trimming. The decision to trim low-quality bases involves a tradeoff between removing unreliable data and preserving read length for alignment.

Adapter Trimming and Read Filtering

Adapter sequences are added during library preparation and may be read when the insert fragment is shorter than the read length. These adapter sequences must be removed before alignment because they do not correspond to the biological sample. Adapter trimming tools identify and remove adapter sequences from reads, and the choice of trimming parameters affects how many reads are retained.

Read filtering decisions include removing reads that are too short after trimming, removing reads with excessive low-quality bases, and removing reads that align to ribosomal RNA or other unwanted sequences. The filtering stringency should balance the need for clean data against the risk of discarding informative reads. Overly aggressive filtering can remove legitimate biological signal, while insufficient filtering can introduce artifacts.

Quality Control Records

Documenting the quality control process is essential for reproducibility. The number of reads before and after each filtering step, the proportion of reads removed at each stage, and the reasons for removal should be recorded. These records allow researchers to assess whether the filtering decisions were appropriate and to compare quality across samples. Systematic differences in read counts or quality metrics between experimental groups can indicate batch effects that need to be addressed in later analysis stages.

Read Alignment to Reference Genome or Transcriptome

Once reads pass quality control, they must be aligned to a reference to determine their genomic origin. The choice between aligning to a reference genome or a reference transcriptome depends on the analysis goals and the availability of a high-quality reference for the organism under study.

Spliced Alignment Algorithms

RNA-seq reads span exon junctions, requiring aligners that can handle spliced alignments. Spliced aligners split reads across exon boundaries and can identify novel splice junctions. The choice of aligner affects alignment speed, memory usage, and accuracy. Published comparisons of HISAT and STAR-based pipelines provide practical information about the performance characteristics of these two popular aligners, and the choice between them often depends on the specific computational environment and data characteristics.

STAR is known for its speed and accuracy but requires substantial memory, while HISAT2 uses a graph-based index that is more memory-efficient. Both aligners can handle spliced alignments and produce output in standard formats. The selection between them should consider the size of the reference genome, the available computational resources, and the need for specialized features such as chimeric alignment detection.

Reference Genome Selection

The reference genome version and annotation determine the coordinates to which reads are aligned and the gene models used for quantification. Different versions of the same reference genome may produce different results, so the reference version must be recorded and used consistently across all samples in a study. For organisms with well-annotated genomes, the reference provides a reliable framework for alignment and quantification.

For organisms without a high-quality reference genome, researchers may need to use a reference transcriptome or perform de novo transcriptome assembly. These approaches have different limitations and require careful validation. The choice of reference has downstream consequences for quantification and differential expression analysis, so it should be made deliberately and documented thoroughly.

Alignment Quality Metrics

After alignment, several metrics indicate whether the alignment process performed well. The overall alignment rate, the proportion of reads mapping to exonic versus intronic regions, and the distribution of reads across genes provide information about library quality and alignment accuracy. Low alignment rates may indicate contamination, adapter problems, or reference mismatches.

The proportion of reads that map to multiple locations in the genome is an important consideration. Multi-mapping reads, which align equally well to multiple genomic locations, are common in gene families with high sequence similarity. The handling of multi-mapping reads affects quantification accuracy, and different quantification tools use different strategies for assigning these reads.

Quantification of Gene and Transcript Expression Levels

Quantification converts aligned reads into expression measurements for genes or transcripts. The choice of quantification approach has a major impact on the resulting expression values and the downstream analysis.

Gene-Level Quantification

Gene-level quantification counts the number of reads that overlap with each gene's genomic coordinates. This approach is straightforward and works well for most differential expression analyses. Tools that count reads overlapping gene features, such as featureCounts, produce count matrices that serve as input to differential expression packages.

The gene model annotation used for counting determines which genomic regions are assigned to each gene. Annotation files from public databases provide the gene coordinates, and the choice of annotation version affects the results. Overlapping genes, alternative isoforms, and genes on opposite strands require careful handling during counting.

Transcript-Level Quantification

Transcript-level quantification estimates the abundance of individual isoforms, which requires resolving reads that map to regions shared by multiple isoforms. This is a more complex problem than gene-level counting because reads must be assigned to specific isoforms based on their exon composition. Tools that perform transcript-level quantification use statistical models to estimate isoform abundances from the observed read patterns.

Transcript-level quantification provides more detailed information than gene-level counts but requires deeper sequencing and is more sensitive to annotation errors. The choice between gene-level and transcript-level quantification depends on whether isoform-specific information is needed for the biological question.

Alignment-Free Quantification

Some quantification tools skip the alignment step entirely and quantify transcript abundance directly from raw reads. These alignment-free methods use k-mer matching to assign reads to transcripts and are substantially faster than alignment-based approaches. The tradeoff is reduced ability to detect novel transcripts or splice junctions.

Alignment-free quantification is particularly useful for large datasets or when computational resources are limited. The results are comparable to alignment-based methods for well-annotated organisms, but the limitations should be understood before choosing this approach.

Normalization of Expression Data

Normalization adjusts raw counts to account for technical differences between samples, such as sequencing depth, library composition, and other systematic biases. Without normalization, differences in sequencing depth between samples would be misinterpreted as biological differences in expression.

Library Size Normalization

The simplest normalization approach scales counts by the total number of reads in each sample. This accounts for differences in sequencing depth but does not address composition effects, where a few highly expressed genes in one sample suppress the apparent expression of other genes. More sophisticated normalization methods account for these composition effects by using a subset of genes or a median of ratios approach.

The choice of normalization method affects the results of differential expression analysis. Different normalization methods make different assumptions about the data, and the appropriate choice depends on the experimental design and the characteristics of the data. The normalization method should be selected based on the specific analysis goals and should be documented in the methods.

Batch Effect Correction

Batch effects are technical variations introduced by processing samples in different groups, such as different sequencing runs or different library preparation dates. These effects can obscure biological differences or create spurious differences between groups. Detecting batch effects requires examining the data for systematic patterns that correlate with technical factors instead of biological factors.

Several approaches exist for correcting batch effects, including including batch as a covariate in the statistical model or using dedicated batch correction tools. The choice of approach depends on whether the batch effect is balanced across experimental groups and whether the correction method is compatible with the downstream analysis. Unbalanced batch effects, where all samples from one condition are in one batch, are particularly problematic because the batch effect cannot be distinguished from the biological effect.

Quality Control Visualization

Visualization plays an important role in assessing the effectiveness of normalization and detecting problems in the data. Principal component analysis and hierarchical clustering reveal whether samples group by biological condition or by technical factors. Samples that cluster by batch instead of by experimental group indicate batch effects that need correction.

These visualizations also identify outlier samples that may need to be removed from the analysis. Outlier detection should be based on objective criteria and documented in the analysis records. Removing outliers after seeing the results can introduce bias, so the criteria for outlier removal should be established before examining the differential expression results.

Differential Expression Analysis

Differential expression analysis identifies genes whose expression differs significantly between experimental conditions. This is the central analytical step for most RNA-seq studies, and the statistical methods used have a major impact on the results.

Statistical Models for Count Data

RNA-seq count data follow a negative binomial distribution, which accounts for the overdispersion observed in biological replicates. Differential expression tools such as DESeq2 and edgeR fit statistical models to the count data and test for significant differences between conditions. These tools estimate the dispersion of each gene's expression across replicates and use this information to calculate test statistics.

The choice between DESeq2 and edgeR depends on the specific features of each tool and the experimental design. Both tools are widely used and produce reliable results when applied appropriately. The SARTools pipeline provides a framework for running both DESeq2 and edgeR with systematic quality control steps and diagnostic plots, which helps prevent errors from misusing the methods.

Multiple Testing Correction

Testing thousands of genes simultaneously creates a multiple testing problem, where the probability of false positives increases with the number of tests performed. Multiple testing correction methods control the false discovery rate, which is the expected proportion of false positives among the genes declared significant. The choice of correction method and the significance threshold affect the balance between sensitivity and specificity.

The false discovery rate is typically controlled at 5 percent, meaning that approximately 5 percent of the genes declared significant are expected to be false positives. More stringent thresholds reduce false positives but also reduce the power to detect true differences. The appropriate threshold depends on the goals of the study and the consequences of false positives versus false negatives.

Effect Size and Biological Significance

Statistical significance does not necessarily imply biological significance. A gene can show a statistically significant difference that is too small to be biologically meaningful. Reporting the effect size, typically as the log2 fold change, alongside the statistical significance provides a more complete picture of the results.

The choice of fold change threshold depends on the biological context and the expected magnitude of expression changes. Some biological processes involve large expression changes, while others involve subtle shifts that are nonetheless biologically important. The thresholds should be justified based on the biological question and documented in the analysis.

Diagnostic Plots and Model Checking

Differential expression analysis should include diagnostic plots that check whether the statistical model fits the data. Dispersion estimates, mean-variance relationships, and the distribution of p-values provide information about whether the analysis assumptions are met. Systematic quality control steps are recommended throughout the analysis process, and failing to check the characteristics of the dataset may lead to spurious results.

The SARTools pipeline generates HTML reports that display diagnostic plots for quality control and model hypothesis checking while keeping track of the whole analysis process, parameter values, and versions of the R packages used. This type of documentation supports reproducibility and helps identify problems in the analysis.

Functional Analysis and Biological Interpretation

The output of differential expression analysis is a list of genes that change between conditions. Functional analysis places these genes in biological context by identifying pathways, processes, and functions that are enriched among the differentially expressed genes.

Enrichment Analysis Methods

Enrichment analysis tests whether specific functional categories, such as Gene Ontology terms or biological pathways, are overrepresented among the differentially expressed genes compared to the background set of all measured genes. The choice of background set affects the results, and the background should reflect the genes that were actually tested for differential expression.

Several databases provide functional annotations for genes, including pathway databases and Gene Ontology. The choice of annotation database depends on the organism and the type of functional information needed. Different databases may produce different enrichment results, so the choice should be documented and justified.

Gene Regulatory Network Inference

For studies focused on regulatory mechanisms, gene regulatory network inference can identify transcription factors that control the expression of differentially expressed genes. The pySCENIC pipeline maps transcription factors onto gene regulatory networks using single-cell RNA-seq data, integrating various cell types to infer cell-specific regulatory networks. This approach identifies potential transcription factor targets based on co-expression, performs motif enrichment analysis to identify direct targets, and scores the activity of regulons on cell types.

Gene regulatory network inference is more complex than standard enrichment analysis and requires careful validation. The inferred networks represent hypotheses about regulatory relationships that should be tested experimentally.

Meta-Analysis of Multiple Datasets

Individual RNA-seq studies often have limited numbers of biological samples, which restricts the statistical power to detect small but meaningful expression changes. Meta-analysis integrates multiple datasets from different origins to maximize the sample size and improve statistical power. A sensibly designed meta-analysis of select individual studies can reveal latent insights that are not apparent from individual data analysis.

Meta-analysis requires careful attention to batch effects and technical differences between studies. The integration of gene expression data from different origins is challenging and encompasses a series of data analysis steps that must be completed before arriving at meaningful biological insights. The potential benefits of increased statistical power must be weighed against the complexities of harmonizing data across studies.

Single-Cell RNA-Seq Analysis Considerations

Single-cell RNA-seq has enabled gene expression to be studied at an unprecedented resolution, but the analysis workflow differs substantially from bulk RNA-seq. The steps of a typical single-cell RNA-seq analysis include preprocessing, quality control, normalization, data correction, feature selection, and dimensionality reduction, followed by cell and gene-level downstream analysis.

Quality Control for Single-Cell Data

Single-cell data present unique quality control challenges because each cell represents an individual library. Cells with low total RNA content, high mitochondrial read fractions, or other quality issues must be identified and filtered. The quality control thresholds for single-cell data depend on the cell type and the experimental protocol.

Dimensionality Reduction and Clustering

Single-cell analysis typically involves reducing the high-dimensional expression data to a lower-dimensional representation that captures the major sources of variation. Principal component analysis and other dimensionality reduction methods prepare the data for clustering, which groups cells by their expression profiles. The clustering resolution determines the number of cell populations identified.

Visual Analytics for Single-Cell Workflows

The complexity of single-cell analysis workflows has driven the development of visual analytics tools that support exploration across the full pipeline. Existing visualization systems typically support only isolated stages or data types, leaving full-pipeline exploration under-addressed. Newer tools aim to provide unified environments for exploration, comparison, and workflow tracking across the single-cell analysis pipeline.

Common Failure Patterns and How to Avoid Them

Several recurring problems undermine RNA-seq analyses. Recognizing these failure patterns helps researchers avoid them and diagnose issues when they arise.

Insufficient Quality Control

Skipping or rushing quality control is the most common cause of unreliable results. Samples with poor quality that pass through to downstream analysis can create spurious differential expression or obscure real differences. The quality control steps should be treated as mandatory, not optional, and the results should be examined carefully before proceeding.

Inappropriate Normalization

Using the wrong normalization method or applying normalization incorrectly can produce misleading results. The normalization method should match the data characteristics and the analysis goals. Failing to check whether normalization adequately removed technical variation can leave batch effects in the data.

Overlooking Batch Effects

Batch effects that are not detected and corrected can produce false positives or false negatives in differential expression analysis. The potential for batch effects should be considered at the experimental design stage, and the data should be examined for batch patterns before proceeding to differential expression.

Misinterpreting Statistical Significance

Equating statistical significance with biological importance leads to overinterpretation of results. The effect sizes should be considered alongside the p-values, and the biological plausibility of the findings should be assessed. Validation of key findings with an independent method, such as RT-qPCR, provides additional confidence in the results.

Poor Documentation

Failure to document analysis steps, software versions, and parameters makes results impossible to reproduce. The analysis records should include the reference genome version, the aligner and parameters, the quantification method, the normalization approach, and the differential expression settings. This documentation supports reproducibility and allows others to assess the analysis quality.

Records and Measurements for Reproducible Analysis

Maintaining detailed records throughout the analysis process is essential for reproducibility and for troubleshooting problems. The following records should be maintained for each analysis:

Analysis Log

The analysis log records each step performed, the software and version used, the parameters selected, and the date of execution. This log provides a complete history of the analysis that can be reviewed and reproduced. Version control systems for analysis scripts help track changes over time.

Quality Metrics Summary

The quality metrics from each stage should be compiled into a summary table that allows comparison across samples. This summary includes read counts before and after filtering, alignment rates, and the proportion of reads assigned to genes. Systematic differences between samples or groups can indicate problems that need investigation.

Session Information

The R session information or equivalent records for other analysis environments document the software versions and dependencies used in the analysis. This information is critical for reproducing the analysis in a different environment or at a later time.

Professional Escalation Criteria

Some analysis situations require consultation with bioinformatics specialists or statisticians. The following situations warrant escalation:

Unexplained Quality Problems

When quality metrics indicate problems that cannot be resolved with standard trimming and filtering approaches, consultation with a specialist may be needed. Persistent low alignment rates, unusual GC content distributions, or unexpected contamination patterns may require specialized troubleshooting.

Complex Experimental Designs

Experimental designs with multiple factors, paired samples, or repeated measures require statistical models that are more complex than standard two-group comparisons. A statistician with experience in genomic data analysis should be consulted to ensure the model is specified correctly.

Conflicting Results Across Methods

When different analysis tools produce substantially different results, the discrepancy should be investigated before drawing conclusions. A specialist can help determine whether the differences reflect genuine biological complexity or methodological issues.

Regulatory or Compliance Requirements

Studies involving human subjects or regulated applications may have specific requirements for data handling, analysis validation, and reporting. The NIH Genomic Data Sharing Policy describes expectations for data sharing and participant protections, and researchers should consult with institutional compliance offices when these requirements apply.

Frequently Asked Questions

What is the difference between alignment to a genome and alignment to a transcriptome?

Alignment to a reference genome maps reads to their genomic coordinates, which allows detection of novel splice junctions and does not depend on existing transcript annotations. Alignment to a reference transcriptome maps reads directly to known transcript sequences, which is faster but cannot detect novel transcripts or splice variants. The choice depends on whether the analysis requires discovery of new transcript structures or only quantification of known genes.

How many biological replicates are needed for reliable differential expression analysis?

The number of biological replicates needed depends on the variability in the system and the magnitude of expression changes to be detected. Three replicates per condition is often considered the practical minimum, while five or more replicates provide substantially better statistical power. The relationship between replicate number and power depends on the specific experimental system, and researchers should consider the expected effect sizes when planning their experiments.

What is the difference between gene-level and transcript-level quantification?

Gene-level quantification counts reads that overlap with any exon of a gene, providing a single expression value per gene. Transcript-level quantification estimates the abundance of individual isoforms, which requires resolving reads that map to regions shared by multiple isoforms. Transcript-level analysis provides more detailed information but requires deeper sequencing and is more sensitive to annotation quality.

Why is normalization necessary in RNA-seq analysis?

Normalization adjusts raw counts to account for technical differences between samples, such as sequencing depth and library composition. Without normalization, samples sequenced more deeply would appear to have higher expression for all genes, and differences in library composition could create spurious expression differences. Normalization makes expression values comparable across samples so that biological differences can be detected.

What is a batch effect and how can it be detected?

A batch effect is technical variation introduced by processing samples in different groups, such as different sequencing runs or different library preparation dates. Batch effects can be detected by examining whether samples cluster by technical factors instead of by biological condition in principal component analysis or hierarchical clustering. Systematic patterns that correlate with processing dates or other technical factors indicate batch effects.

How should multiple testing correction be applied in differential expression analysis?

Multiple testing correction controls the false discovery rate when testing thousands of genes simultaneously. The false discovery rate is typically controlled at 5 percent, meaning that about 5 percent of the genes declared significant are expected to be false positives. The choice of correction method and threshold should balance the need to detect true differences against the risk of false positives.

What is the role of functional enrichment analysis after differential expression?

Functional enrichment analysis identifies biological pathways, processes, and functions that are overrepresented among differentially expressed genes. This analysis places the differentially expressed genes in biological context and generates hypotheses about the mechanisms underlying the observed expression changes. The choice of annotation database and background gene set affects the results and should be documented.

How does single-cell RNA-seq analysis differ from bulk RNA-seq analysis?

Single-cell RNA-seq analysis includes additional steps for filtering low-quality cells, normalizing across cells with different RNA content, reducing dimensionality, and clustering cells by expression profile. The analysis focuses on identifying cell populations and characterizing their gene expression patterns instead of comparing average expression between conditions. The computational demands are substantially higher for single-cell data due to the large number of individual cell libraries.

Related Bioinformatics Guides

• Alternative Splicing Analysis from RNA-Seq Data
• RNA-Seq Differential Expression: DESeq2, edgeR, and limma-voom Frameworks
• Single-Cell RNA-Seq Analysis Pipelines for Veterinary Immunology
• RNA Seq Fastq Example: Structural Analysis and Computational Methodologies in Bioinformatics
• Single-Cell RNA-Seq Normalization: Batch Effect Correction and Dimension Reduction (PCA, t-SNE, UMAP)

References and Further Reading

• EMBL-EBI Training. European Bioinformatics Institute.
• NCBI Data Resources. National Center for Biotechnology Information.
• Genomic Data Sharing Policy. National Institutes of Health.
• The FAIR Guiding Principles. Scientific Data.
• A survey of best practices for RNA-seq data analysis.. Genome biology, 2016.
• Current best practices in single-cell RNA-seq analysis: a tutorial.. Molecular systems biology, 2019.
• An RNA-Seq Data Analysis Pipeline.. Methods in molecular biology (Clifton, N.J.), 2024.
• Inference of Gene Regulatory Network from Single-Cell Transcriptomic Data Using pySCENIC.. Methods in molecular biology (Clifton, N.J.), 2021.
• Improved meta-analysis pipeline ameliorates distinctive gene regulators of diabetic vasculopathy in human endothelial cell (hECs) RNA-Seq data.. PloS one, 2023.
• RNA-seq: Basic Bioinformatics Analysis.. Current protocols in molecular biology, 2018.
• RNA-seq Data Analysis for Differential Expression.. Methods in molecular biology (Clifton, N.J.), 2022.
• Sequences to Differences in Gene Expression: Analysis of RNA-Seq Data.. Methods in molecular biology (Clifton, N.J.), 2022.
• Designing scFlowVis: Visual Analytics for Single-Cell RNA Sequencing Analysis.. 2026.
• Stepwise Protocol for Alternative Splicing Analysis in Single-Cell SMART-Seq2 RNA-Seq Data. 2026.
• LongPolyASE: an end-to-end framework for allele-specific gene and isoform analysis in polyploids using long-read RNA-seq.. 2026.
• An Agentic AI Framework for Ingestion and Standardization of Single-Cell RNA-seq Data Analysis. bioRxiv, 2025.
• SARTools: A DESeq2- and EdgeR-Based R Pipeline for Comprehensive Differential Analysis of RNA-Seq Data. bioRxiv, 2015.
• Development and validation of the PipeSeq program for RNA-seq data analysis in the Chlamydomonas reinhardtii as a model. Vavilovskii zhurnal genetiki i selektsii, 2026.
• Improving the flexibility of RNA-Seq data analysis pipelines. Proceedings IEEE International Workshop on Genomic Signal Processing and Statistics, 2012.
• Comparing HISAT and STAR-based pipelines for RNA-Seq Data Analysis: A real experience. Proceedings IEEE Symposium on Computer Based Medical Systems, 2023.
• Seqacademy: An educational pipeline for RNA-seq and chip-seq analysis.. F1000research, 2018.
• SeqAcademy: An educational pipeline for RNA-Seq and ChIP-Seq analysis. F1000research, 2020.

This article is educational and does not replace validated analysis plans, institutional policy, clinical interpretation, or specialist review.