RNA-Seq Alignment Tools: STAR, HISAT2, and Beyond

RNA-seq alignment is the computational step that determines where each sequenced read originated in the genome or transcriptome. The choice of aligner affects downstream quantification, splice junction detection, and differential expression results. This article compares STAR, HISAT2, Salmon, and related tools for researchers who need to select an aligner appropriate for their experimental design, computational resources, and biological questions.

At a Glance

The table below summarizes key characteristics of commonly used RNA-seq alignment and quantification tools. Performance metrics vary by dataset, read length, and computing environment, so treat these values as general guidance instead of fixed specifications.

STAR was developed to address the challenge of aligning large RNA-seq datasets with high accuracy and speed. The original publication reported mapping 550 million paired-end reads per hour on a 12-core server while improving alignment sensitivity and precision compared to existing tools. STAR uses a maximum mappable seed search approach followed by seed clustering and stitching. It can discover non-canonical splices and chimeric fusion transcripts, and it is implemented as standalone open source C++ code.

HISAT2 represents the next generation of the HISAT aligner family and uses a graph-based approach for fast splice-aware alignment. In cloud-based RNA-seq analysis comparisons, HISAT2 was identified as the fastest of the classical aligners with good alignment quality. It is commonly paired with StringTie for transcript assembly and with DESeq2 or edgeR for differential expression analysis.

Salmon and Kallisto use pseudoalignment or lightweight alignment approaches that map reads to transcriptomes instead of full genomes. These tools are substantially faster than genomic aligners and are well suited for gene expression quantification when splice junction discovery is not the primary objective. In cloud infrastructure benchmarks, pseudoalignment algorithms such as Kallisto and Salmon combined high read quality estimation with cost-efficient runtime performance.

Understanding RNA-seq Alignment Fundamentals

RNA-seq reads present unique alignment challenges compared to DNA sequencing reads. The primary complication is that RNA transcripts undergo splicing, so a single read may span two or more exons that are distant in the genomic sequence. Splice-aware aligners must account for this non-contiguous transcript structure when mapping reads to a reference genome.

The alignment problem is further complicated by the massive scale of RNA-seq data. Modern sequencing platforms generate hundreds of millions of short reads per experiment, and the alignment step is often the computational bottleneck in RNA-seq analysis pipelines. The choice of aligner must balance accuracy against computational cost, and the optimal balance depends on the specific research question.

Splice-Aware Versus Transcriptome Alignment

Splice-aware aligners map reads to a reference genome while allowing for intron-spanning alignments. These tools can detect both annotated and novel splice junctions, making them essential for studies of alternative splicing, novel isoform discovery, and fusion transcript detection. STAR and HISAT2 are the most widely used splice-aware aligners.

Transcriptome alignment tools map reads directly to a reference transcriptome instead of a genome. This approach is computationally efficient because reads are aligned contiguously to transcript sequences without the need to consider intronic regions. Salmon and Kallisto use variations of this approach, and they are particularly effective for gene and transcript quantification when the transcriptome annotation is complete and accurate.

The choice between genomic and transcriptome alignment depends on the analysis goals. If the objective is to quantify expression of known genes and transcripts, transcriptome-based methods offer speed and accuracy. If the objective includes discovering novel splice junctions, identifying fusion transcripts, or analyzing repetitive element expression, splice-aware genomic alignment is necessary.

Read Length and Sequencing Platform Considerations

STAR can align spliced sequences of any length with moderate error rates, providing scalability for emerging sequencing technologies. This flexibility makes STAR suitable for both short-read Illumina data and longer reads from platforms such as PacBio or Oxford Nanopore.

For Ion Torrent single-end RNA-seq samples, benchmarking studies found that CLC and BWA-MEM achieved the best results in terms of efficiency, accuracy, duplication rate, saturation profile, and running time. For Illumina paired-end transcriptomics data, Novoalign excelled in accuracy and saturation analyses. These findings illustrate that no single aligner performs best across all sequencing platforms and experimental designs.

Core Principles for Aligner Selection

Selecting an RNA-seq aligner requires evaluating several factors that interact with the specific characteristics of your dataset and research objectives. The following principles guide the selection process.

Accuracy Requirements

Alignment accuracy directly affects all downstream analyses. In a comparison of HISAT2 and STAR using breast cancer progression samples, STAR generated more precise alignments, particularly for early neoplasia samples, while HISAT2 was prone to misaligning reads to retrogene genomic loci. This finding demonstrates that aligner choice can introduce systematic biases that affect biological conclusions.

For splice junction detection, a benchmark on simulated human data found that DeepSAP achieved the highest mean F1-score for splice junction detection, outperforming DRAGEN, novoSplice, STAR, HISAT2, and Subjunc. DeepSAP integrates transcriptome-guided genomic alignment with transformer-based splice junction scoring, representing an emerging approach that combines classical alignment with machine learning.

Computational Resource Constraints

Memory usage and runtime are practical considerations that can determine whether an aligner is feasible for a given computing environment. In single-cell RNA-seq comparisons, STAR produced more genes and higher gene expression values than Kallisto, but this gain in accuracy came with significantly slower computation time (4-fold) and higher memory usage (7.7-fold) compared to Kallisto.

For researchers working with limited computational resources, HISAT2 offers a favorable balance. In a comparison of aligners using grapevine powdery mildew RNA-seq data, HISAT2 was approximately 3-fold faster than the next fastest aligner in runtime. The same study found that all aligners performed well in terms of alignment rate and gene coverage, with the exception of TopHat2, which HISAT2 superseded.

Downstream Analysis Compatibility

The aligner output format and the information preserved in alignment files determine compatibility with downstream tools. STAR generates output files that can be used for transcript and gene expression quantification, differential gene expression analysis, novel isoform reconstruction, and signal visualization. HISAT2 output is commonly used with StringTie for transcript assembly and with featureCounts for read counting.

For integrated analysis of RNA-seq and ATAC-seq data, platforms such as RAGER automate workflows that combine popular bioinformatics tools. RAGER uses Snakemake to provide an automated pipeline that minimizes the need for bioinformatics expertise and reduces processing time. Similarly, the PipeSeq pipeline combines HISAT2 alignment with StringTie assembly, featureCounts counting, and DESeq2 differential expression analysis in an automated workflow.

Practical Workflow for Aligner Implementation

Implementing an RNA-seq alignment workflow requires attention to data preparation, parameter selection, and quality assessment. The following steps outline a practical approach.

Step 1: Assess Input Data Quality

Before alignment, evaluate the quality of raw sequencing reads. Check base quality scores, adapter contamination, and GC bias. While most aligners can handle reads with moderate quality issues, excessive adapter contamination or low-quality bases can reduce alignment accuracy and introduce bias in downstream quantification.

For repetitive element analysis, note that transposable elements can contribute a substantial portion of the cell transcriptome but are typically ignored in most RNA-seq analyses. Properly aligning short reads to these highly repetitive regions is complex, and tools such as TEtranscripts are designed to assign both uniquely and ambiguously mapped reads to all possible gene and transposable element transcripts.

Step 2: Prepare Reference Files

Download the appropriate reference genome and annotation files from authoritative sources such as the NCBI Data Resources. Ensure that the genome version and annotation version are compatible with each other and appropriate for your organism.

Generate the genome index for your chosen aligner. STAR requires genome index generation before alignment, and this step can be memory intensive. HISAT2 uses a graph-based index that includes known splice sites and genetic variants, which reduces memory requirements compared to STAR.

Step 3: Configure Alignment Parameters

STAR alignment can be controlled by many user-defined parameters, and the choice of parameters affects both mapping accuracy and speed. Key parameters include the number of mismatches allowed, the minimum and maximum intron sizes, and the handling of multi-mapping reads.

For most applications, default parameters provide a reasonable starting point. However, optimization may be necessary for specific experimental designs. For example, single-cell RNA-seq data may benefit from parameters that account for the lower complexity of individual cell libraries.

Step 4: Run Alignment and Monitor Performance

Execute the alignment and monitor runtime, memory usage, and alignment statistics. STAR reports the percentage of uniquely mapped reads, multi-mapped reads, and unmapped reads. These statistics provide an initial indication of alignment quality.

For large datasets, consider using cloud-based or high-performance computing resources. STAR has been optimized for high-throughput computing in cloud environments, and pseudoalignment tools such as Salmon and Kallisto are particularly well suited for cost-efficient cloud deployment.

Step 5: Assess Alignment Quality

Evaluate alignment quality using multiple metrics. The percentage of uniquely mapped reads should be high for most RNA-seq datasets, typically above 70% for well-annotated organisms. Low unique mapping rates may indicate sample contamination, poor reference quality, or excessive repetitive content.

Check the distribution of reads across gene features. A high percentage of reads mapping to exonic regions indicates good library quality, while excessive intronic or intergenic mapping may suggest genomic DNA contamination or annotation issues.

Step 6: Document and Archive Analysis Parameters

Record all software versions, reference genome versions, and parameter settings used in the alignment. This documentation is essential for reproducibility and for interpreting results in the context of the analysis pipeline.

Options and Tradeoffs in Aligner Choice

The aligner landscape includes tools with fundamentally different approaches to read mapping. Understanding the tradeoffs between these approaches helps researchers make informed decisions.

STAR for Comprehensive Genomic Alignment

STAR is designed for accurate spliced sequence alignment at ultrafast speed. It performs unbiased de novo detection of canonical junctions and can discover non-canonical splices and chimeric fusion transcripts. STAR can also map full-length RNA sequences, making it suitable for a wide range of applications.

The primary limitation of STAR is its memory footprint. For the human genome, STAR requires approximately 30 GB of RAM for genome indexing and alignment. This requirement can be prohibitive for researchers working with standard desktop computers or limited cloud instances.

STAR is particularly well suited for studies that require comprehensive splice junction detection, fusion transcript identification, or analysis of complex RNA sequence arrangements such as circular RNA. It is also the aligner of choice for many single-cell RNA-seq pipelines, where its higher gene detection rates and accuracy justify the increased computational cost.

HISAT2 for Fast and Memory-Efficient Alignment

HISAT2 uses a hierarchical graph-based index that represents the genome and known genetic variants. This approach enables fast alignment with lower memory requirements than STAR. HISAT2 is the fastest of the classical aligners with good alignment quality, making it suitable for large datasets and resource-constrained environments.

HISAT2 is commonly used in automated pipelines. The PipeSeq pipeline uses HISAT2 for read alignment to the reference genome, followed by StringTie for transcript assembly, featureCounts for transcript counting, and DESeq2 for statistical analysis of differential gene expression. This pipeline has been validated using Chlamydomonas reinhardtii RNA-seq data, with results verified by RT-qPCR for 12 GATA transcription factor genes.

One limitation of HISAT2 is its tendency to misalign reads to retrogene genomic loci in some sample types. This issue was observed in breast cancer FFPE samples, where STAR generated more precise alignments. Researchers working with samples that may contain processed pseudogenes or retrogenes should be aware of this potential bias.

Salmon and Kallisto for Transcript Quantification

Salmon and Kallisto use pseudoalignment or lightweight alignment approaches that map reads to transcriptomes instead of genomes. These tools are substantially faster than genomic aligners and require less memory, making them attractive for large-scale studies and cloud-based analysis.

In single-cell RNA-seq comparisons, Kallisto showed lower gene detection and gene expression values compared to STAR. However, the computational savings were substantial, with STAR requiring 4-fold more computation time and 7.7-fold more memory than Kallisto. For studies where computational resources are limited and the focus is on quantifying known transcripts, Salmon or Kallisto may be appropriate.

Salmon and Kallisto are also well suited for cloud-based analysis. In benchmarks of transcript quantification methods for cloud deployment, pseudoalignment algorithms such as Kallisto and Salmon combined high read quality estimation with cost-efficient runtime performance.

Emerging Tools and Approaches

DeepSAP represents an emerging approach that integrates GSNAP transcriptome-guided genomic alignment with transformer-based splice junction scoring. This tool improves splice junction detection, indel identification, and resolution of complex splicing patterns. On the Baruzzo human simulated benchmark, DeepSAP achieved the highest mean F1-score for splice junction detection, outperforming DRAGEN, novoSplice, STAR, HISAT2, and Subjunc.

StringTie3 is a major update to the StringTie assembler, specifically designed for total RNA-seq. It introduces a nascent mode that models co-transcriptional splicing to separate nascent from mature transcripts, and a refined long-read module that distinguishes genuine polyadenylation sites from poly(A)-priming artifacts. StringTie3 substantially reduces assembly errors and outperforms existing tools across short-read, long-read, and hybrid-read datasets.

Observations and Measurements for Aligner Evaluation

Evaluating aligner performance requires systematic measurement of multiple metrics. The following observations from published benchmarks provide context for interpreting aligner performance.

Alignment Rate and Gene Coverage

In a comparison of Bowtie2, BWA, HISAT2, MUMmer4, STAR, and TopHat2 using RNA-seq data from grapevine powdery mildew, all aligners performed well in terms of alignment rate and gene coverage, with the exception of TopHat2. BWA had the best performance in these metrics for most transcripts, while HISAT2 and STAR performed well for longer transcripts over 500 base pairs.

This finding illustrates that aligner performance can vary by transcript length and genomic context. Researchers should consider the characteristics of their target transcripts when selecting an aligner.

Gene Detection Sensitivity

In single-cell RNA-seq comparisons, STAR globally produced more genes and higher gene expression values compared to Kallisto and Bowtie2. STAR also yielded higher correlations of the Gini index for genes with RNA-FISH validation results. Using 10x Genomics PBMC 3K single-cell RNA-seq and mouse cortex single nuclei RNA-seq data, STAR showed similar or better cell-type annotation results by detecting a larger subset of known gene markers.

The higher gene detection sensitivity of STAR is particularly important for single-cell studies, where dropout events and low expression levels make sensitive detection critical for identifying cell types and states.

Differential Expression Concordance

The choice of aligner can affect differential expression results. In a comparison of HISAT2 and STAR using breast cancer progression samples, edgeR and DESeq2 produced similar lists of differentially expressed genes, with edgeR producing more conservative though shorter lists. Gene Ontology enrichment analysis revealed no skewness in significant GO terms identified among differentially expressed genes by edgeR versus DESeq2.

This finding suggests that while aligner choice can affect individual gene measurements, the impact on downstream differential expression results may be moderated by the statistical methods used. However, the observed differences in aligner performance for specific sample types, such as FFPE samples, indicate that aligner choice remains an important consideration for clinical research.

Records and Documentation for Reproducible Analysis

Reproducibility in RNA-seq analysis requires careful documentation of all computational steps. The following practices support reproducible and interpretable results.

Version Control and Parameter Logging

Record the exact versions of all software used in the analysis pipeline, including the aligner, reference genome, and annotation files. Document all non-default parameters and the rationale for parameter choices. This information should be included in the methods section of publications and in analysis reports.

Alignment Statistics Tracking

Maintain a record of alignment statistics for each sample, including total reads, uniquely mapped reads, multi-mapped reads, and unmapped reads. These statistics provide a quality metric that can be compared across samples and experiments. Sudden changes in alignment rates may indicate sample quality issues or technical problems.

Data Management and Sharing

Follow established data sharing policies when working with human genomic data. The NIH Genomic Data Sharing Policy provides requirements for data sharing, privacy protection, and data security. For other organisms, consider depositing raw sequencing data and processed alignment files in public repositories such as the NCBI Sequence Read Archive.

Adherence to the FAIR Guiding Principles supports findability, accessibility, interoperability, and reusability of research data. These principles apply to both raw sequencing data and processed alignment results.

Common Failure Patterns and Troubleshooting

Several recurring problems can compromise RNA-seq alignment. Recognizing these patterns helps researchers diagnose and correct issues.

Low Unique Mapping Rates

Low percentages of uniquely mapped reads can result from sample contamination, poor reference quality, or excessive repetitive content. For transposable element analysis, note that nearly half of the human genome is derived from transposable element sequences, and up to two thirds of the human genome might be repeat-derived. Reads originating from these regions are inherently difficult to map uniquely.

If low unique mapping rates are observed, check for adapter contamination, ribosomal RNA contamination, and sample cross-contamination. Consider whether the reference genome includes all relevant sequences, including alternate haplotypes and unplaced contigs.

Excessive Memory Usage

STAR alignment requires substantial memory, particularly for large genomes. If memory usage exceeds available resources, consider using HISAT2, which has lower memory requirements, or optimize STAR parameters to reduce memory usage. Cloud-based implementations of STAR have been optimized for high-throughput computing environments.

Misalignment to Retrogene Loci

HISAT2 has been observed to misalign reads to retrogene genomic loci in some sample types. This issue can affect gene expression quantification for genes with processed pseudogenes. If this problem is suspected, compare results with an alternative aligner such as STAR to assess the impact on biological conclusions.

Inconsistent Results Across Aligners

Different aligners can produce different results for the same dataset. In a comparison of RNA-seq and microarray data analysis, the choice of analysis method affected results. When aligner choice substantially affects biological conclusions, consider validating key findings with an orthogonal approach such as RT-qPCR.

Limitations and Interpretation Boundaries

RNA-seq alignment tools have inherent limitations that affect the interpretation of results. Understanding these boundaries is essential for drawing appropriate biological conclusions.

Multi-Mapped Reads

Reads that map to multiple genomic locations present a challenge for all aligners. These reads are particularly common in repetitive regions and gene families with high sequence similarity. Different aligners handle multi-mapped reads differently, and the choice of approach can affect quantification results.

Tools such as TEtranscripts are designed to assign both uniquely and ambiguously mapped reads to all possible gene and transposable element transcripts in order to statistically infer the correct abundances. This approach is appropriate when repetitive element expression is of interest.

Annotation Dependence

Transcriptome-based quantification tools such as Salmon and Kallisto depend on the completeness and accuracy of transcriptome annotations. If the annotation is incomplete, reads from novel transcripts will not be quantified correctly. Splice-aware genomic aligners such as STAR and HISAT2 can detect novel transcripts but require additional assembly steps to reconstruct them.

Platform-Specific Performance

Aligner performance varies by sequencing platform. For Ion Torrent single-end RNA-seq samples, CLC and BWA-MEM were the most suitable aligners in terms of efficiency, accuracy, duplication rate, saturation profile, and running time. For Illumina paired-end data, Novoalign excelled in accuracy and saturation analyses. Researchers should consider the sequencing platform when selecting an aligner.

Safety and Regulatory Context

RNA-seq analysis involving human samples is subject to regulatory requirements that affect data management and sharing. The NIH Genomic Data Sharing Policy establishes expectations for the sharing of genomic data generated through NIH-funded research. This policy addresses data submission, data sharing timelines, and privacy protections for human data.

Researchers working with human RNA-seq data should ensure compliance with applicable regulations and institutional review board requirements. Data sharing plans should be developed at the study design stage, and data management practices should align with the FAIR Guiding Principles.

For clinical research applications, the choice of bioinformatics tools becomes critical. In breast cancer research using FFPE samples, STAR and edgeR were identified as well-suited tools for differential gene expression analysis. The choice of aligner can affect the reliability of clinical research findings, and validation of key results with orthogonal methods is recommended.

Professional Escalation Criteria

Certain situations warrant consultation with bioinformatics specialists or escalation to more experienced colleagues. Consider seeking expert assistance in the following circumstances.

Complex Experimental Designs

Experiments involving multiple conditions, time points, or treatment groups may require specialized analysis approaches. If the experimental design includes batch effects, confounding variables, or repeated measures, consult a bioinformatics specialist to ensure appropriate analysis methods.

Unusual Alignment Patterns

If alignment statistics deviate substantially from expected values, or if results differ dramatically between aligners, seek expert advice before proceeding with downstream analysis. Unexpected patterns may indicate technical issues, sample problems, or biological phenomena that require specialized analysis.

Clinical or Regulatory Applications

RNA-seq analysis for clinical decision-making or regulatory submissions requires rigorous validation and documentation. Consult with bioinformatics specialists and regulatory experts to ensure that analysis methods meet applicable standards and requirements.

Novel or Non-Model Organisms

Alignment to non-model organism genomes can be challenging due to incomplete annotations and assembly quality issues. If the reference genome is fragmented or poorly annotated, consult with specialists who have experience with similar organisms.

Frequently Asked Questions

What is the difference between splice-aware alignment and pseudoalignment?

Splice-aware alignment maps reads to a reference genome while allowing reads to span exon-exon junctions. This approach can detect novel splice junctions and fusion transcripts. Pseudoalignment maps reads to a reference transcriptome without full base-by-base alignment, which is faster but limited to known transcripts. STAR and HISAT2 are splice-aware aligners, while Salmon and Kallisto use pseudoalignment approaches.

How much memory does STAR require for human RNA-seq data?

STAR requires approximately 30 GB of RAM for human genome indexing and alignment. This memory requirement can be prohibitive for standard desktop computers. HISAT2 has lower memory requirements and may be more suitable for resource-constrained environments. Cloud-based implementations of STAR have been optimized for high-throughput computing.

Can Salmon or Kallisto replace STAR for differential expression analysis?

Salmon and Kallisto can be used for differential expression analysis when the focus is on quantifying known transcripts. These tools are substantially faster and require less memory than STAR. However, STAR detects more genes and produces higher gene expression values in single-cell data, and it can discover novel splice junctions and fusion transcripts. The choice depends on whether novel transcript discovery is needed.

Why does HISAT2 misalign reads to retrogene loci?

HISAT2 has been observed to misalign reads to retrogene genomic loci in some sample types, particularly in FFPE breast cancer samples. This issue may arise from the graph-based index structure that includes known genetic variants. STAR generated more precise alignments in these samples. If retrogene misalignment is suspected, compare results with an alternative aligner.

What alignment statistics should I report in my methods section?

Report the aligner name and version, reference genome version and source, annotation version, and all non-default parameters. Include alignment statistics such as total reads, uniquely mapped reads, multi-mapped reads, and unmapped reads. Document the computing environment including operating system, CPU, and memory specifications.

How do I choose between STAR and HISAT2 for my experiment?

Consider the size of your dataset, available computational resources, and analysis goals. STAR provides high accuracy and comprehensive splice junction detection but requires substantial memory. HISAT2 is faster and uses less memory but may misalign reads to retrogene loci in some samples. For large datasets with limited resources, HISAT2 may be more practical. For comprehensive transcript discovery, STAR is preferred.

What is the role of transcriptome assembly tools in RNA-seq analysis?

Transcriptome assembly tools such as StringTie, Cufflinks, and Scallop reconstruct full-length transcripts from aligned reads. These tools are used in conjunction with splice-aware aligners such as HISAT2 and STAR. StringTie3 is a major update designed for total RNA-seq that separates nascent from mature transcripts and reduces assembly errors.

How does aligner choice affect single-cell RNA-seq analysis?

Aligner choice significantly affects single-cell RNA-seq results. STAR produces more genes and higher gene expression values compared to Kallisto, and it detects a larger subset of known gene markers for cell-type annotation. However, STAR requires 4-fold more computation time and 7.7-fold more memory than Kallisto. The choice depends on whether sensitivity or computational efficiency is prioritized.

Related Bioinformatics Guides

• Alternative Splicing Analysis from RNA-Seq Data
• Single-Cell RNA-Seq Analysis Pipelines for Veterinary Immunology
• Single-cell RNA-seq Trajectory Inference and Cell Lineage Tracing
• Single-Cell RNA-seq Clustering and Cell-Type Annotation Pipelines
• RNA Seq Fastq Example: Structural Analysis and Computational Methodologies in Bioinformatics

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.
• STAR: ultrafast universal RNA-seq aligner.. Bioinformatics (Oxford, England), 2013.
• Mapping RNA-seq Reads with STAR.. Current protocols in bioinformatics, 2015.
• Interleukin 11 therapy causes acute left ventricular dysfunction.. Cardiovascular research, 2024.
• Optimizing RNA-Seq Mapping with STAR.. Methods in molecular biology (Clifton, N.J.), 2016.
• Evaluation of STAR and Kallisto on Single Cell RNA-Seq Data Alignment.. G3 (Bethesda, Md.), 2020.
• Mapping RNA-seq reads to transcriptomes efficiently based on learning to hash method.. Computers in biology and medicine, 2020.
• Analysis of RNA-Seq Data Using TEtranscripts.. Methods in molecular biology (Clifton, N.J.), 2018.
• Protocol for transcriptome assembly by the TransBorrow algorithm.. Biology methods & protocols, 2023.
• DeepSAP: improved RNA-seq alignment by integrating transcriptome guidance with transformer-based splice junction scoring.. 2026.
• RAGER: A user-friendly computational platform for integrated analysis of RNA-Seq and ATAC-seq data.. 2026.
• Development and validation of the PipeSeq program for RNA-seq data analysis in the Chlamydomonas reinhardtii as a model.. 2026.
• DiRT v2.0: An Optimized Pipeline for Detecting Dicistronic tRNA-mRNA Transcripts in Plants. 2026.
• Transcriptomic profiling of the central amygdala in a rat model of diabetes-associated neuropathic pain.. 2026.
• StringTie3 improves total RNA-seq assembly by resolving nascent and mature transcripts.. 2026.
• New evaluation methods of read mapping by 17 aligners on simulated and empirical NGS data: an updated comparison of DNA- and RNA-Seq data from Illumina and Ion Torrent technologies. Neural computing & applications (Print), 2021.
• Aligning the Aligners: Comparison of RNA Sequencing Data Alignment and Gene Expression Quantification Tools for Clinical Breast Cancer Research. Journal of Personalized Medicine, 2019.
• Qualitative Comparison of Selected Indel Detection Methods for RNA-Seq Data. International Work-Conference on Bioinformatics and Biomedical Engineering, 2019.
• Interoperable RNA-Seq analysis in the cloud.. Biochimica et Biophysica Acta. Gene Regulatory Mechanisms, 2020.
• Comparison of Short-Read Sequence Aligners Indicates Strengths and Weaknesses for Biologists to Consider. Frontiers in Plant Science, 2021.
• Benchmarking RNA-Seq Aligners at Base-Level and Junction Base-Level Resolution Using the Arabidopsis thaliana Genome. Plants, 2024.
• Optimizing Star Aligner for High Throughput Computing in the Cloud. Proceedings 2024 IEEE International Conference on Cluster Computing Workshops Cluster Workshops 2024, 2024.
• A comparative study of RNA-Seq and microarray data analysis on the two examples of rectal-cancer patients and Burkitt Lymphoma cells. Plos One, 2018.
• Performance evaluation of lossy quality compression algorithms for RNA-seq data. BMC Bioinformatics, 2020.

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