Sequencing Pipeline: From Raw Data to Variants

Direct Answer

A sequencing pipeline converts raw instrument output into interpretable variant calls through a defined sequence of computational steps: quality control, read alignment, post-alignment processing, variant calling, filtering, and annotation. For targeted sequencing applications in diagnostic and research settings, the pipeline must be validated, reproducible, and documented so that results can be compared across runs and laboratories. This article explains each pipeline stage, the decisions required at each step, the records you should keep, and the common failure points that compromise variant accuracy.

The reader is assumed to be a laboratory student, technician, researcher, or diagnostic professional who has sequencing data ready for analysis or who needs to establish a pipeline for a new assay. The focus is targeted sequencing, where a defined set of genomic regions is enriched before sequencing, but the principles extend to whole-genome, RNA sequencing, and metagenomic applications.

At a Glance

The table below summarizes the main pipeline stages, the primary purpose of each stage, and the key output that moves to the next step.

Each stage has specific parameters that must be recorded and justified. The pipeline is not a single fixed script but a series of decisions that depend on the sequencing platform, the target region design, the question being asked, and the acceptable error rate for the intended use.

Scope and Context for Targeted Sequencing Analysis

Targeted sequencing enriches specific genomic regions before sequencing, which reduces cost and increases depth at loci of interest. The bioinformatics pipeline for targeted data must account for the enrichment method, the amplicon or capture design, and the expected uniformity of coverage across target regions.

The choice of analysis approach depends on the biological question. RNA sequencing analysis differs from DNA variant calling because it must account for splicing, transcript abundance, and the possibility that a variant is present in only some transcripts. A review of RNA-seq best practices notes that no single analysis pipeline fits all cases, and each major step from experimental design through differential expression carries specific challenges. The same principle applies to DNA pipelines: the pipeline must be matched to the assay and the question.

For diagnostic applications, the pipeline must be stable and reproducible. A case study of variant calling pipeline selection in molecular diagnostics showed that the choice of variant search protocol can affect the genetic diagnostics outcome for a specific patient. The authors raised the question of whether synthetic performance metrics such as precision and recall matter more than stability and reproducibility when a pipeline change could alter a treatment decision. This underscores the need to validate any pipeline change against known reference materials before it is used for patient samples.

The pipeline should be treated as part of the assay system. Just as you would validate a new reagent lot or a new instrument, you should validate a new pipeline version or a new alignment tool before it is used for reporting.

Core Principles of Sequencing Data Analysis

Raw Data Is Not Ready for Interpretation

Sequencing instruments produce base calls with associated quality scores. These raw reads contain adapter sequences, low-quality bases, and possibly contamination from other organisms or from the sequencing platform itself. The first principle of pipeline design is that raw data must be assessed before any biological interpretation is attempted.

Quality control serves two purposes. First, it determines whether the sequencing run met the expected metrics for yield, quality, and coverage. Second, it identifies problems that can be corrected, such as adapter contamination or poor-quality base calls at read ends. A pipeline that skips quality control risks propagating errors into every downstream step.

Reference Dependence

Most variant calling pipelines align reads to a reference genome. The choice of reference build matters because variant coordinates, gene annotations, and population frequency databases are tied to specific reference versions. If you change the reference build, you must re-annotate your variant calls and re-validate the pipeline.

For RNA sequencing, alignment can be to a reference genome or to a transcriptome, and the choice affects how splice junctions are handled. The alignment step is pivotal for decoding the origins and functions of the sequences, and errors at this stage propagate into quantification and differential expression results.

Reproducibility Requires Version Control

Every tool in the pipeline has a version, and every parameter has a value. Reproducibility requires that you record both. A pipeline that runs today with one version of an aligner may produce different results with a newer version. The cloud-based pipeline described for nanopore sequencing data analysis included ongoing maintenance and testing of version changes before implementation, which is a model for any diagnostic pipeline.

Validation Against Known Truth

A pipeline should be tested against samples with known variants. This can be a reference cell line with a published variant set, a previously characterized clinical sample, or a synthetic construct with introduced variants. The validation establishes the sensitivity and specificity of the pipeline for the intended variant types and allele frequencies.

Quality Control of Raw Sequencing Data

What Quality Control Measures

Quality control assesses the raw FASTQ files before alignment. The main metrics are per-base quality scores, per-sequence quality scores, GC content, adapter contamination, duplicate rates, and the number of reads that pass quality thresholds.

Per-base quality scores show whether quality degrades toward the ends of reads, which is common with certain sequencing chemistries. Adapter contamination appears when the insert size is shorter than the read length, causing the sequencer to read into the adapter sequence. High duplicate rates can indicate low library complexity or excessive PCR amplification during library preparation.

Quality Control Decisions

The quality control step produces a decision: proceed with the data as is, trim low-quality bases and adapters, or reject the run. The decision thresholds should be defined in advance and recorded. For example, you might define a minimum mean quality score per read, a maximum adapter contamination rate, and a minimum number of reads per sample.

Trimming is a common corrective action. Reads are truncated at positions where quality drops below a threshold, and adapter sequences are removed. The trimming parameters must be recorded because they affect downstream alignment and variant calling.

Quality Control Records

For each sample, record the following:

• Total number of raw reads
• Number and percentage of reads passing quality filters
• Mean quality score before and after trimming
• Adapter contamination rate
• GC content and any deviation from expected
• Trimming parameters used

These records allow you to compare runs over time and to identify systematic problems with library preparation or sequencing.

Read Alignment

Alignment Principles

Read alignment maps each read to its most likely position in the reference genome. The aligner must handle mismatches caused by sequencing errors and true variants, insertions and deletions, and for RNA data, reads that span splice junctions.

The choice of aligner depends on the data type. DNA aligners are optimized for contiguous alignment, while RNA aligners must account for spliced reads. The RNA-seq analysis pipeline described in the literature uses tools such as HISAT2 for alignment, followed by HTSeq for quantification, demonstrating that the aligner choice is tied to the downstream analysis steps.

Alignment Parameters

Key alignment parameters include:

• Maximum number of mismatches allowed
• Whether to allow gaps and how many
• How to handle reads that map to multiple locations
• Whether to use known splice junctions for RNA data

These parameters affect sensitivity and specificity. A strict mismatch threshold may miss true variants, while a lenient threshold may increase false alignments. The parameters should be chosen based on the expected sequence diversity in the target regions and validated with known samples.

Alignment Output and Metrics

The alignment output is a BAM or CRAM file containing the read alignments. Alignment metrics to record include:

• Percentage of reads mapped
• Percentage of reads mapped to the target regions
• Mean coverage depth across targets
• Percentage of target bases covered at a defined depth
• Percentage of reads with multiple alignments

For targeted sequencing, the percentage of reads on target is a critical metric. Low on-target rates indicate problems with the enrichment step or with the reference used for alignment.

Post-Alignment Processing

Duplicate Removal

PCR duplicates arise when the same library molecule is amplified and sequenced multiple times. Duplicate reads can inflate coverage estimates and create false variant calls if the duplication is biased. Marking or removing duplicates is standard practice for DNA sequencing, particularly for amplicon-based targeted panels where PCR amplification is extensive.

The duplicate rate should be recorded. A high duplicate rate reduces the effective sequencing depth and may indicate too many PCR cycles during library preparation.

Base Quality Recalibration

Base quality scores from the sequencer may be systematically inaccurate. Recalibration adjusts the quality scores based on empirical error rates observed in the data. This step improves the accuracy of variant calling, particularly for low-frequency variants.

Recalibration requires known variant sites or a set of high-confidence calls to estimate error rates. For targeted panels, the recalibration can be performed using the known polymorphisms in the target regions.

Realignment Around Indels

Insertions and deletions are more difficult to align than single nucleotide variants. Reads spanning an indel may be misaligned near the variant, causing false mismatches. Local realignment around known or suspected indel sites improves the accuracy of indel calling.

Modern variant callers often perform realignment internally, but the decision to use an external realignment step should be documented.

Variant Calling

Variant Caller Selection

The variant caller identifies positions where the sample differs from the reference. The choice of caller affects the results. Different callers use different statistical models, and they may disagree on low-confidence variants.

For DNA data, common callers include HaplotypeCaller, which reassembles the active region to identify variants. For RNA-seq data, variant calling is complicated by the absence of a matched normal sample in many cases. A refined variant calling pipeline for RNA-seq data from breast cancer cell lines used HaplotypeCaller and then applied filtering to exclude low read depth sites, RNA-edit sites, and low complexity regions. Common variants were filtered using population databases. This example shows that the filtering strategy is as important as the caller itself.

Variant Calling Parameters

Key parameters include:

• Minimum base quality at the variant position
• Minimum read depth at the variant position
• Minimum allele frequency for calling a variant
• Ploidy of the sample

For targeted sequencing, the expected allele frequency depends on the sample type. Germline variants are expected at 50 percent or 100 percent allele frequency, while somatic variants may be present at much lower frequencies. The pipeline must be configured for the expected allele frequencies.

Variant Call Format Output

The variant caller produces a VCF file with the raw variant calls. Each variant has a quality score, read depth, allele frequency, and genotype. The raw VCF should not be interpreted directly. It must be filtered and annotated before reporting.

Variant Filtering

Why Filtering Is Necessary

Raw variant calls include artifacts from sequencing errors, alignment errors, and mapping to repetitive regions. Filtering removes these artifacts while retaining true variants. The filtering strategy must balance sensitivity and specificity.

A case study of variant calling in breast cancer cell lines without matched-normal samples illustrates the scale of filtering. Starting from hundreds of thousands of single nucleotide variants and small insertions and deletions, about a thousand variants remained after filtering for each sample. The filtering excluded low read depth sites, RNA-edit sites, and low complexity regions, and removed common variants using population databases. This example demonstrates that filtering is not a minor step but a major reduction in the number of candidate variants.

Filtering Criteria

Common filtering criteria include:

• Read depth at the variant site
• Quality score from the variant caller
• Allele frequency
• Strand bias, which indicates whether the variant is supported by reads from only one strand
• Mapping quality
• Presence in repetitive or low complexity regions

The thresholds for each criterion should be defined based on the validation data. If the pipeline is validated with known variants, the thresholds can be adjusted to maximize detection of true variants while minimizing false calls.

Population Frequency Filtering

For germline variant calling, variants that are common in the general population are unlikely to be pathogenic. Population frequency databases such as gnomAD and dbSNP are used to filter common variants. The frequency threshold depends on the disease and the mode of inheritance. For rare disease diagnostics, a low frequency threshold is used.

For somatic variant calling without a matched normal, population frequency filtering is essential to remove common genetic variation. The breast cancer cell line study used 1000 Genomes, gnomAD, and dbSNP data to filter common variants.

Variant Annotation

Annotation Content

Annotation adds biological context to the filtered variant list. The annotation includes:

• Gene name and transcript identifier
• Variant type and effect on the protein
• Amino acid change
• Clinical significance if known
• Population frequency
• Conservation scores

Annotation databases are updated regularly, and the version must be recorded. A variant that is classified as of unknown significance today may be reclassified as pathogenic when new evidence is published.

Annotation Tools

Annotation tools use transcript databases to determine the effect of each variant. The choice of transcript set matters because different transcripts may have different variant effects. The annotation should be reviewed by a person with domain expertise before the variant is reported.

Annotation Records

Record the annotation database version, the transcript set used, and the date of annotation. This information is necessary for re-annotation if the databases change.

RNA Sequencing Pipeline Considerations

Differences from DNA Pipelines

RNA sequencing analysis differs from DNA variant calling in several ways. The reads represent transcribed RNA, so the depth at a given position reflects gene expression level instead of genomic copy number. Splicing means that reads may not align contiguously to the genome. The analysis pipeline must account for these differences.

An RNA-seq analysis pipeline typically includes quality control, preprocessing, alignment, quantification of gene and transcript levels, and differential expression analysis. The pipeline described in the literature is divided into three sections: initial data download and quality control, gene-level quantification, and differential expression analysis. This structure reflects the primary questions asked of RNA-seq data.

Quantification and Differential Expression

Quantification assigns reads to genes or transcripts. The quantification method affects the results, and the choice depends on whether the analysis is at the gene level or the transcript level. Differential expression analysis compares expression levels between conditions and identifies genes with significant differences.

The RNA-seq analysis pipeline described in the literature includes functional analysis of differentially expressed genes using Gene Ontology and pathway analyses. This step contextualizes the roles of the identified genes within broader biological frameworks.

Variant Calling from RNA-Seq Data

Variant calling from RNA-seq data is possible but has limitations. The depth at a given position depends on expression level, so lowly expressed genes may have insufficient depth for confident variant calls. RNA editing can create apparent variants that are not present in the DNA. The breast cancer cell line study found that missing variants could mainly be attributed to sites with low read depth.

If RNA-seq data is used for variant calling, the limitations must be documented and the results should be confirmed with DNA sequencing when clinically significant.

Long-Read Sequencing Pipelines

Platform Differences

Long-read sequencing platforms produce reads that are much longer than short-read platforms. The analysis pipeline differs in alignment and variant calling because the reads can span complex structural variants and repetitive regions that are difficult to analyze with short reads.

Third-generation sequencing technologies have improved read length and base-calling accuracy while reducing costs per base. These technologies show potential in difficult-to-sequence genomic regions, structural variation detection, and RNA expression profiling. However, protocol standardization and the development of easy-to-use pipelines for data analysis are needed for routine diagnostic applications.

Pipeline Adaptation

Long-read pipelines must handle higher error rates in the raw reads, which are often corrected during analysis. The alignment and variant calling tools are different from those used for short reads. The cloud-based nanopore pipeline described in the literature used Docker containers to include all tools in the computational environment and compared genotype concordance between the manual method and the automated pipeline to assess data accuracy.

Metagenomic Sequencing Pipelines

Purpose and Challenges

Metagenomic sequencing analyzes DNA from mixed microbial communities. The pipeline must identify the organisms present and their relative abundances. The large amount of data produced by metagenomic sequencing requires comprehensive pipelines that automate the analytical steps.

The HOME-BIO pipeline for metagenomic shotgun sequencing data analysis comprises three independent analytical modules for inclusive analysis of large datasets. It removes low-complexity and problematic reads, integrates analytical steps for a comprehensive taxonomy profile of each sample, and is customizable according to user needs.

Pipeline Components

Metagenomic pipelines include quality control, removal of host reads, taxonomic classification, and functional analysis. The taxonomic classification can be performed by aligning reads to reference databases or by using k-mer-based methods. The choice of database affects the results, and the database version must be recorded.

Workflow Management and Automation

Why Workflow Systems Are Used

The rapidly increasing amounts of sequencing data have made data processing infeasible without automated pipelines. Workflow management systems integrate data and analytic resources and simplify the task of data analysis. They allow users to create multi-step analyses, modify parameters, and run the pipeline in a reproducible manner.

The Closha workflow system provides fast and cost-effective analysis of massive genomic data using high-performance computing clusters. It allows users to create multi-step analyses using drag and drop functionality and to modify the parameters of pipeline tools. Users can also import Galaxy pipelines into Closha.

Cloud-Based Pipelines

Cloud computing offers on-demand access to computing resources without the need to purchase hardware. The nanopore sequencing pipeline using Microsoft Azure demonstrated that cloud-based analysis can meet the key imperatives for performance, cost, usability, simplicity, and accuracy. The pipeline allowed for ongoing maintenance and testing of version changes before implementation.

Containerization

Containerization packages all tools and dependencies into a single image that can be run on any system with the container runtime. This ensures that the pipeline runs with the same tool versions regardless of the host system. The nanopore pipeline used Docker containers to include all tools in the cloud computational environment.

Pipeline Validation and Quality Assurance

Validation with Known Samples

A pipeline must be validated before it is used for reporting. Validation uses samples with known variants to measure sensitivity and specificity. The validation samples should represent the range of variant types and allele frequencies expected in the intended use.

The proficiency test for bioinformatics analysis of salmonid rhabdovirus genome sequences illustrates the importance of validation. Three partners analyzed 75 raw datasets using their own bioinformatics pipelines, and the results were compared. The overall reproducibility was 99.94 percent, but 526 discrepancies were observed, with 46.4 percent located in coding regions. Manual curation was identified as the most critical step affecting sequence comparability.

Quality Control in the Pipeline

Quality control should be built into the pipeline at multiple stages, also at the beginning. Alignment metrics, coverage metrics, and variant call metrics should be reviewed for each sample. A sample with low coverage or an unexpected variant rate should be flagged for review.

Documentation Requirements

The Laboratory Quality Management System Handbook from the World Health Organization provides guidance on the documentation and quality assurance practices expected in a laboratory setting. The handbook emphasizes the importance of standard operating procedures, records, and quality control for reliable laboratory results.

For a sequencing pipeline, documentation should include:

• Pipeline version and tool versions
• Parameter values for each step
• Reference genome version and annotation database versions
• Validation results
• Quality control metrics for each run

Records and Measurements

What to Record for Each Run

For each sequencing run and each sample, maintain the following records:

• Sample identifier and source
• Library preparation method and kit lot number
• Sequencing platform and reagent lot numbers
• Raw read count and quality metrics
• Alignment metrics including on-target rate and mean depth
• Duplicate rate
• Variant call count before and after filtering
• Filtering thresholds applied
• Annotation database versions
• Final variant list and any variants flagged for review

Run-Level Metrics

Run-level metrics allow you to monitor pipeline performance over time. Track the mean quality score, the percentage of reads passing filters, the on-target rate, and the mean coverage depth. A gradual decline in any metric may indicate a problem with reagents, the instrument, or the pipeline.

Sample-Level Metrics

Sample-level metrics identify individual samples that may have failed. A sample with unusually low coverage, a high duplicate rate, or an unexpected variant count should be investigated. The investigation may reveal a problem with the sample, the library preparation, or the sequencing run.

Common Failure Patterns

Low On-Target Rate

A low percentage of reads mapping to the target regions indicates a problem with the enrichment step or with the reference used for alignment. Possible causes include failed hybridization or amplification, contamination with non-target DNA, or a mismatch between the target design and the reference genome.

High Duplicate Rate

A high duplicate rate reduces the effective sequencing depth and can create false variant calls. Possible causes include too many PCR cycles during library preparation, low input DNA, or over-sequencing of the library.

Strand Bias in Variant Calls

Strand bias occurs when a variant is supported by reads from only one strand. This pattern is often associated with sequencing artifacts. Variant callers report strand bias metrics, and variants with significant strand bias should be reviewed carefully.

Coverage Gaps in Target Regions

Some target regions may have consistently low coverage. This can be caused by difficult-to-sequence regions such as GC-rich or repetitive sequences. If the gaps are consistent across samples, the target design may need to be revised.

Discordance Between Pipeline Versions

Changing a tool version or a parameter value can change the variant calls. The case study of variant calling pipeline selection in molecular diagnostics demonstrated that the choice of variant search protocol can affect the genetic diagnostics outcome. Any pipeline change should be validated with known samples before it is used for reporting.

Limitations and Interpretation Boundaries

Depth Limitations

Variant calling requires sufficient read depth at the variant position. Low depth sites are unreliable, and the breast cancer cell line study found that missing variants could mainly be attributed to sites with low read depth. The minimum depth for confident variant calling depends on the allele frequency and the error rate.

Allele Frequency Limitations

The ability to detect a variant depends on its allele frequency. Germline variants at 50 percent allele frequency are easier to detect than somatic variants at 5 percent allele frequency. The pipeline must be validated for the allele frequencies expected in the intended use.

Reference Genome Limitations

The reference genome is not complete or perfect. Some regions are difficult to align, and some variants in repetitive regions cannot be called reliably. These limitations should be documented in the interpretation of results.

RNA-Seq Specific Limitations

Variant calling from RNA-seq data has additional limitations. The depth depends on expression level, RNA editing can create apparent variants, and the absence of a matched normal sample complicates filtering. These limitations should be documented when RNA-seq data is used for variant calling.

Safety and Regulatory Context

Biosafety Considerations

The Laboratory Biosafety Manual from the World Health Organization provides guidance on the safe handling of biological materials. While the bioinformatics pipeline itself does not involve handling biological materials, the samples that produce the sequencing data may contain infectious agents. Laboratory staff should follow the biosafety practices appropriate for the sample types being processed.

Quality Management

The Laboratory Quality Management System Handbook from the World Health Organization describes the quality assurance practices expected in a laboratory. These practices apply to the bioinformatics pipeline as much as to the wet laboratory steps. The pipeline should have standard operating procedures, records, and quality control measures.

Regulatory Considerations for Diagnostics

For diagnostic applications, the pipeline may be subject to regulatory oversight. The Bioanalytical Method Validation Guidance from the U.S. Food and Drug Administration describes the validation expectations for bioanalytical methods. While this guidance is not specific to sequencing pipelines, the principles of validation, documentation, and quality control apply.

The Assay Guidance Manual from the National Center for Advancing Translational Sciences provides guidance on assay development and validation. The manual emphasizes the importance of understanding the assay performance characteristics and documenting the validation.

Professional Escalation Criteria

When to Escalate a Pipeline Issue

Escalate to a supervisor or a bioinformatics specialist when:

• The quality control metrics fall outside the defined acceptance criteria
• The alignment metrics indicate a systematic problem
• The variant calls are discordant with expected results
• A pipeline change produces different results on validation samples
• The results are needed for a clinical decision and the pipeline has not been validated for the sample type

When to Escalate a Variant Interpretation Issue

Escalate to a molecular geneticist or clinical specialist when:

• A variant is identified in a gene associated with the clinical indication
• The variant is novel and has no known clinical significance
• The variant is in a region with known alignment difficulties
• The variant is present at a low allele frequency and the clinical significance is uncertain

Documentation of Escalation

Record the reason for escalation, the person to whom the issue was escalated, and the outcome. This documentation is part of the quality management system and supports the traceability of decisions.

Practical Implementation Steps

Step 1: Define the Pipeline Requirements

Define the intended use of the pipeline, the sample types, the variant types of interest, and the acceptable error rates. These requirements guide the choice of tools and parameters.

Step 2: Select Tools and Parameters

Select the tools for each pipeline stage based on the requirements and the available evidence. Record the tool versions and parameter values.

Step 3: Validate the Pipeline

Validate the pipeline with known samples. Measure sensitivity and specificity for the variant types and allele frequencies of interest. Document the validation results.

Step 4: Establish Quality Control Metrics

Define the quality control metrics and acceptance criteria for each pipeline stage. These metrics should be monitored for every run.

Step 5: Document the Pipeline

Write a standard operating procedure that describes each pipeline stage, the tools and parameters, the quality control metrics, and the escalation criteria. The procedure should be reviewed and approved.

Step 6: Train Staff

Train laboratory staff on the pipeline procedure, the quality control metrics, and the escalation criteria. Document the training.

Step 7: Monitor and Review

Monitor the pipeline metrics over time and review the pipeline periodically. Update the pipeline when tools or databases change, and re-validate after any change.

Frequently Asked Questions

What is the difference between primary, secondary, and tertiary bioinformatics analysis?

Primary analysis converts raw instrument signals into base calls and quality scores. Secondary analysis includes quality control, alignment, and variant calling. Tertiary analysis includes filtering, annotation, and interpretation. The boundaries between these stages vary by laboratory, but the distinction is useful for organizing the pipeline and assigning responsibilities.

How do I choose between different variant callers?

The choice of variant caller depends on the data type, the variant types of interest, and the validation results. Different callers use different statistical models and may disagree on low-confidence variants. Validate each caller with known samples and choose the one that meets the sensitivity and specificity requirements for the intended use.

Why does my pipeline produce different results after a tool update?

Tool updates can change alignment algorithms, variant calling models, or default parameters. Any change can affect the results. Validate the updated pipeline with known samples before using it for reporting, and document the validation.

How much read depth is needed for confident variant calling?

The required depth depends on the allele frequency of the variants of interest and the error rate of the sequencing platform. Higher depth is needed for low-frequency variants. The minimum depth should be determined during validation and documented in the pipeline procedure.

Can I call variants from RNA-seq data?

Variant calling from RNA-seq data is possible but has limitations. The depth at a given position depends on expression level, RNA editing can create apparent variants, and the absence of a matched normal sample complicates filtering. Confirm clinically significant variants with DNA sequencing.

What is the role of population frequency databases in variant filtering?

Population frequency databases such as gnomAD and dbSNP are used to filter common variants that are unlikely to be pathogenic. The frequency threshold depends on the disease and the mode of inheritance. The database version must be recorded because the content changes over time.

How do I handle samples with low coverage in target regions?

Low coverage in target regions reduces confidence in variant calls. Possible responses include resequencing the sample, using a different enrichment method, or reporting the region as not covered. The decision should be documented and the limitations should be noted in the interpretation.

What should I do if my pipeline produces discordant results on the same sample?

Discordant results can arise from changes in the pipeline, differences in sample quality, or random variation. Investigate the cause by reviewing the quality control metrics and the variant calls. If the discordance is due to a pipeline change, re-validate the pipeline with known samples.

Related Diagnostic Guides

• Procedure for Quality Control: Step-by-Step Implementation in a Molecular Lab
• How to Perform a Gram Stain: Protocol and Quality Control
• How to Interpret DNA Sequencing Chromatograms: Peaks, Quality, and Heterozygotes
• Quality Control Analysis: Methods for Monitoring Lab Performance
• Gram Staining Protocol: Reagents, Procedure, and Quality Control

References and Further Reading

• Laboratory Quality Management System Handbook. World Health Organization.
• Laboratory Biosafety Manual. World Health Organization.
• Assay Guidance Manual. National Center for Advancing Translational Sciences.
• Bioanalytical Method Validation Guidance. U.S. Food and Drug Administration.
• NCBI Literature Resources. National Center for Biotechnology Information.
• A survey of best practices for RNA-seq data analysis.. Genome biology, 2016.
• RNA-Seq Experiment and Data Analysis.. Methods in molecular biology (Clifton, N.J.), 2022.
• RNA-Seq Data Analysis.. Methods in molecular biology (Clifton, N.J.), 2024.
• An RNA-Seq Data Analysis Pipeline.. Methods in molecular biology (Clifton, N.J.), 2024.
• Nanopore sequencing data analysis using Microsoft Azure cloud computing service.. PloS one, 2022.
• The Third-Generation Sequencing Challenge: Novel Insights for the Omic Sciences.. Biomolecules, 2024.
• HOME-BIO (sHOtgun MEtagenomic analysis of BIOlogical entities): a specific and comprehensive pipeline for metagenomic shotgun sequencing data analysis.. BMC bioinformatics, 2021.
• lentiMPRA and MPRAflow for high-throughput functional characterization of gene regulatory elements.. Nature protocols, 2020.
• Refined variant calling pipeline on RNA-seq data of breast cancer cell lines without matched-normal samples.. 2025.
• Case report: A case study of variant calling pipeline selection effect on the molecular diagnostics outcome.. 2024.
• Sequencing of animal viruses: quality data assurance for NGS bioinformatics. Virology Journal, 2019.
• Closha: bioinformatics workflow system for the analysis of massive sequencing data. BMC Bioinformatics, 2018.
• NGS-μsat: bioinformatics framework supporting high throughput microsatellite genotyping from next generation sequencing platforms. Conservation Genetics Resources, 2021.
• Bioinformatics Pathway Analysis Pipeline for NGS Transcriptome Profile Data on Nasopharyngeal Carcinoma. IOP Conference Series: Earth and Environment, 2021.
• Towards multi-approaches bioinformatics pipeline based on big data and cloud computing for next generation sequencing data analysis. Advances in Intelligent Systems and Computing, 2020.
• Streamlining Next-Generation Sequencing Data Analysis with Nextflow and nf-core Pipelines. Next Generation Sequencing Standard Operating Procedures and Applications, 2025.
• Wg-blimp: An end-to-end analysis pipeline for whole genome bisulfite sequencing data. BMC Bioinformatics, 2020.
• [Automatic analysis pipeline of next-generation sequencing data].. Yi Chuan Hereditas Zhongguo Yi Chuan Xue Hui Bian Ji, 2014.

This article is educational and does not replace validated laboratory procedures, institutional biosafety review, manufacturer instructions, or professional interpretation.