A Comprehensive Guide to Metagenomic Read Preprocessing: From Raw FASTQ to Clean Reads

By Dr. Zubair Khalid, DVM, MS, PhD · 2026-08-20

Key Takeaways

• Metagenomic read preprocessing is a critical multi-step computational workflow transforming raw FASTQ sequencing data into clean, validated reads for downstream analysis, directly impacting the biological interpretability of taxonomic profiling, functional annotation, and genome assembly.
• Core preprocessing stages include data integrity validation, quality score interpretation using tools like FastQC, adapter and primer trimming (e.g., with Cutadapt), quality filtering based on per-base scores, and host contamination removal via alignment to reference genomes.
• For paired-end amplicon sequencing, read merging is essential to create contiguous sequences, while shotgun metagenomics often requires more aggressive host filtering due to the presence of host DNA, with both benefiting from workflow management systems like nf-core for reproducibility.
• Reproducibility is paramount, necessitating meticulous documentation of tool versions, parameter settings, and reference database versions, often achieved through containerization (Docker) and version control (Git) to ensure consistent and auditable analysis.
• Common failure patterns include overly aggressive filtering leading to data loss, insufficient adapter trimming causing contamination, host filtering erroneously removing microbial reads, and inconsistent preprocessing across samples, all of which require careful troubleshooting based on quality metrics and documented parameters.

Metagenomic read preprocessing is the sequence of computational steps that convert raw sequencing output in FASTQ format into cleaned, validated reads suitable for downstream analysis such as taxonomic profiling, functional annotation, and genome assembly. This article defines the complete preprocessing workflow, explains the purpose of each stage, and provides practical decision criteria for researchers managing shotgun metagenomic or amplicon datasets. The guidance applies to laboratory professionals, biology students, and life-science practitioners who need a structured approach to handling high-throughput sequencing data before any biological interpretation begins.

The preprocessing phase determines the quality of every subsequent analysis step. Errors introduced or overlooked during this stage propagate through taxonomic classification, abundance estimation, and assembly, producing results that may appear statistically valid but are biologically misleading. Understanding what each preprocessing tool does, why the order of operations matters, and how to document decisions is essential for reproducible microbiome research.

Scope of Metagenomic Read Preprocessing

Metagenomic read preprocessing covers all operations applied to raw sequencing reads between the moment they leave the sequencer and the point where they enter analysis pipelines for taxonomic assignment, functional profiling, or genome binning. This scope includes format validation, quality assessment, adapter trimming, quality filtering, host contamination removal, and in some workflows, error correction and read merging for paired-end data.

The scope excludes downstream analytical steps such as taxonomic classification, functional annotation, assembly, binning, and statistical analysis. However, preprocessing decisions directly influence these downstream outcomes, so the workflow must be designed with the intended analysis in mind. For example, a dataset destined for reference-based taxonomic profiling may require different preprocessing choices than one destined for de novo assembly.

Shotgun metagenomics and amplicon sequencing (such as 16S rRNA metabarcoding) share core preprocessing principles but differ in specific requirements. Shotgun metagenomic data represents the entire genetic content of a microbial community and requires host filtering and often more aggressive quality control. Amplicon data targets specific marker genes and requires primer trimming and often read merging before quality filtering. Both approaches benefit from standardized workflows that are documented and reproducible.

The practical outcome of preprocessing is a set of clean reads with known quality characteristics, accompanied by records that allow another researcher to understand exactly what was done and why. This documentation is a core component of reproducible bioinformatics practice, supported by training resources from organizations such as the Galaxy Training Network and The Carpentries, which emphasize transparent and repeatable analysis methods.

Core Principles of Read Preprocessing

Data Integrity and Format Validation

Raw sequencing data arrives in FASTQ format, which encodes nucleotide sequence information along with per-base quality scores. The first preprocessing step is validating that the files are complete, correctly formatted, and contain the expected number of reads. Corrupted or truncated files can produce silent failures in downstream tools, wasting computational time and producing misleading results.

Validation includes checking that sequence and quality lines have equal length, that quality scores fall within the expected range for the sequencing platform, and that read identifiers are unique and properly formatted. The NCBI maintains documentation on sequence data formats and database submission requirements that provide useful reference points for understanding expected data structures.

For paired-end sequencing, validation must also confirm that forward and reverse read files contain the same number of reads in the same order. Mismatched pairing is a common source of downstream errors that can be difficult to diagnose once analysis has progressed.

Quality Score Interpretation

Quality scores in FASTQ files encode the probability that a given base call is incorrect. Understanding the scoring scheme used by the sequencing platform is necessary for setting appropriate quality thresholds. Different Illumina platforms and software versions have used different encoding schemes, and misinterpreting the encoding can lead to either overly permissive or overly stringent filtering.

Quality assessment tools generate per-base quality distributions, per-sequence quality scores, GC content distributions, and other summary statistics. These outputs allow researchers to identify systematic issues such as quality degradation toward read ends, adapter contamination, and sequencing artifacts. The review of bioinformatics tools for microbial genomics published in Infection, Genetics and Evolution identifies FastQC as a standard tool for this quality assessment stage.

The Role of Reference Databases

Preprocessing decisions often depend on reference databases, particularly for host contamination removal and for amplicon taxonomic assignment. Reference databases are incomplete and biased toward well-studied organisms, which means that reads from novel or underrepresented taxa may fail to match references and be incorrectly removed or classified.

This limitation is documented in the microbial genomics review, which notes that incomplete and biased reference databases remain a persistent challenge in the field. Researchers should document which reference database version was used, when it was downloaded, and how it was prepared, because database updates can substantially change preprocessing outcomes.

Reproducibility as a Core Requirement

Reproducible preprocessing requires that the same inputs and parameters produce the same outputs regardless of when or where the analysis runs. This requirement has driven the development of workflow management systems and containerized pipelines. The nf-core documentation describes community standards for pipeline development that emphasize reproducibility, portability, and documentation.

Containerization using Docker or similar technologies encapsulates software dependencies so that tools run identically across different computing environments. The IMP pipeline for integrated metagenomic and metatranscriptomic analysis demonstrates this approach, using Python and Docker to provide a reproducible and modular analysis framework.

Version control for both code and parameters is essential. Recording the exact tool versions, parameter settings, and reference database versions allows the analysis to be repeated or audited. The Carpentries lessons provide foundational training in version control with Git, which is a practical skill for managing analysis code and documentation.

The Preprocessing Workflow

Step 1: Raw Data Acquisition and Organization

Raw sequencing data should be organized in a consistent directory structure before preprocessing begins. This structure should separate raw data from processed data, document sample metadata, and preserve the original files without modification. The NCBI provides guidance on sequence data organization and submission that can inform local data management practices.

Sample metadata should include information about the biological source, sequencing platform, library preparation method, and any barcode or index sequences used for multiplexing. This metadata is essential for interpreting preprocessing results and for downstream analysis.

Storage considerations are practical and important. Raw sequencing files are large, and the computational bottlenecks and storage capacity limitations noted in the microbial genomics review are real constraints for many laboratories. Planning for data storage, backup, and archival before beginning analysis prevents interruptions and data loss.

Step 2: Quality Assessment

Quality assessment produces summary statistics and visualizations that describe the overall quality of the sequencing run. This step is performed before any filtering or trimming so that the original data quality is documented.

Per-base quality plots show the distribution of quality scores at each position in the read. These plots typically reveal lower quality toward the 3-prime end of reads, which is expected for Illumina sequencing. The rate of quality decline varies between runs and platforms, and this variation informs trimming decisions.

Per-sequence quality scores identify reads that are uniformly poor quality and may represent failed sequencing reactions or contamination. GC content distributions can reveal contamination from other organisms or amplification bias. Overrepresented sequences often indicate adapter contamination or highly abundant organisms.

The Galaxy Training Network provides accessible tutorials on quality assessment and the interpretation of quality reports, making this step approachable for researchers without extensive bioinformatics experience.

Step 3: Adapter and Primer Trimming

Adapter sequences are added during library preparation and must be removed before analysis because they do not represent biological sequence. Adapter contamination is common when insert sizes are shorter than read lengths, causing the sequencer to read through the insert into the adapter.

Trimming tools identify adapter sequences at read ends and remove them. The bioinformatics strategy for 16S and 23S rRNA metabarcoding published in Biotech describes a pipeline that integrates adapter trimming using Cutadapt, which is a widely used tool for this purpose.

For amplicon data, primer sequences must also be removed. Primers are the oligonucleotides used to amplify the target region, and they appear at the start of reads. Failure to remove primers leads to incorrect taxonomic assignment because primer sequences do not match the target gene.

Trimming parameters must balance removing contamination against removing legitimate biological sequence. Overly aggressive trimming can shorten reads to the point where they become unusable for downstream analysis, while insufficient trimming leaves contamination that interferes with classification and assembly.

Step 4: Quality Filtering

Quality filtering removes reads or read segments that do not meet minimum quality thresholds. This step is distinct from trimming, which removes specific sequences such as adapters. Quality filtering is based on the quality scores themselves.

Common approaches include removing reads with average quality below a threshold, trimming read ends until quality exceeds a threshold, and removing reads containing ambiguous bases (N characters). The specific thresholds depend on the downstream analysis and the sequencing platform.

The comparative analysis of HiSeq3000 and BGISEQ-500 sequencing platforms published in Genomics and Informatics demonstrates that different sequencing platforms produce data with different quality characteristics. Platform-specific quality profiles should inform filtering parameters, and researchers should not assume that parameters optimized for one platform transfer directly to another.

For paired-end data, quality filtering can be applied before or after read merging. The order affects how many reads survive and the quality of the merged sequences. The 16S and 23S metabarcoding pipeline described in Biotech performs read merging before quality filtering, which allows quality assessment of the merged product.

Step 5: Host Contamination Removal

Shotgun metagenomic samples often contain host DNA from the organism being studied. For example, gut microbiome samples contain host intestinal cells, and mosquito samples contain mosquito DNA. This host DNA must be removed before analysis because it is not part of the microbial community being studied.

Host filtering works by aligning reads against the host reference genome and removing matching reads. The exploratory metaviromic analysis of the sea-rock pool mosquito Aedes mariae published in Biology highlights the importance of host genome filtering for capturing the diversity of mosquito-associated viruses, noting that sequence similarity between viral and host genomes can complicate this step.

The effectiveness of host filtering depends on the quality and completeness of the host reference genome. Incomplete host genomes leave host reads in the dataset, while overly permissive alignment parameters can remove microbial reads that share sequence similarity with the host.

For non-model organisms without a reference genome, host filtering is more challenging. Options include using closely related genomes, assembling the host genome from the data, or accepting that some host contamination will remain and accounting for it in downstream analysis.

Step 6: Error Correction and Read Merging

Error correction identifies and corrects sequencing errors before downstream analysis. This step is optional but can improve results for certain applications. Error correction algorithms use read overlap and coverage information to identify likely errors and correct them.

For paired-end amplicon data, read merging combines forward and reverse reads into a single contiguous sequence covering the entire amplicon. The 16S and 23S metabarcoding pipeline described in Biotech includes read merging as a critical processing step, followed by quality filtering of the merged product.

Merging requires that the forward and reverse reads overlap sufficiently. The overlap length and the number of mismatches in the overlap region determine whether merging succeeds. Reads that fail to merge may be discarded or retained as unmerged pairs, depending on the analysis requirements.

For shotgun metagenomic data, read merging is less commonly used because the insert sizes are typically longer than the read length, and reads do not overlap. Error correction for shotgun data relies on coverage information from the entire dataset instead of read pairs.

Step 7: Additional Filtering Steps

Depending on the analysis goals, additional filtering steps may be required. These can include removing reads that match known contaminants such as reagents or laboratory contaminants, removing duplicate reads that may represent amplification artifacts, and removing reads with unusual GC content that may indicate contamination.

For virome analysis, the minireview on navigating prokaryotic viral genome analysis from metagenomic data published in mSystems describes preprocessing steps specific to viral datasets, including considerations for the high diversity and methodological biases in viral metagenomics.

For metatranscriptomic data, preprocessing may include steps to remove ribosomal RNA reads, which are highly abundant and not informative for gene expression analysis. The IMP pipeline for integrated metagenomic and metatranscriptomic analysis incorporates robust read preprocessing as part of its workflow.

The decision to apply additional filtering steps should be documented and justified. Each filtering step removes data, and removing too much data can reduce statistical power for detecting rare community members.

At a Glance: Preprocessing Decision Table

Tool Categories and Selection Criteria

General Purpose Quality Control Tools

General purpose quality control tools provide quality assessment, adapter trimming, and quality filtering in a single package. These tools are appropriate for initial preprocessing and for datasets that do not require specialized filtering.

Selection criteria include the tool's ability to handle the specific sequencing platform output, the quality of documentation, the availability of containerized versions, and the tool's performance on datasets of similar size and complexity. The microbial genomics review identifies FastQC as a standard quality assessment tool, and the 16S and 23S metabarcoding pipeline uses VSEARCH and Cutadapt for processing steps.

Amplicon Specific Tools

Amplicon analysis requires tools that handle primer trimming, read merging, and often chimera detection. Chimeras are hybrid sequences formed during PCR amplification when partially extended products serve as primers in subsequent cycles. The 16S and 23S metabarcoding pipeline described in Biotech includes chimera removal as a critical processing step.

Amplicon specific tools often integrate preprocessing with downstream steps such as operational taxonomic unit (OTU) clustering and taxonomic assignment. This integration can simplify the workflow but may reduce flexibility for researchers who want to customize preprocessing parameters.

Shotgun Metagenomic Tools

Shotgun metagenomic preprocessing tools must handle larger datasets and often include host filtering and quality filtering optimized for complex microbial communities. These tools may also include steps for removing low-complexity sequences and for handling reads from highly repetitive genomic regions.

The choice between general purpose tools and specialized shotgun metagenomic tools depends on the analysis goals and the computational resources available. Specialized tools may provide better performance for specific applications but may require more expertise to configure correctly.

Workflow Management Systems

Workflow management systems orchestrate preprocessing steps and manage dependencies between tools. These systems provide reproducibility through version control, parameter documentation, and containerized execution.

The nf-core documentation describes community standards for workflow development that emphasize reproducibility and portability. The IMP pipeline demonstrates the use of workflow management for integrated metagenomic and metatranscriptomic analysis, providing a modular and reproducible framework.

Workflow management systems are particularly valuable for preprocessing because the steps are well defined and the parameters should be consistent across samples within a study. Using a workflow system ensures that all samples receive identical preprocessing, which is essential for valid comparative analysis.

Practical Implementation Steps

Step 1: Define Analysis Goals and Requirements

Before beginning preprocessing, define the downstream analysis goals and the requirements those goals place on read quality and length. Taxonomic profiling with reference-based methods may tolerate shorter reads than de novo assembly. Functional annotation may require higher quality than taxonomic profiling.

Document the sequencing platform, library preparation method, and any known issues from the sequencing run. This information guides parameter selection and helps interpret quality assessment results.

Step 2: Establish the Computational Environment

Set up the computational environment with the required software tools and reference databases. Use containerized versions of tools where available to ensure reproducibility. Record the versions of all software and databases used.

The Bioconductor project provides documentation on package installation and reproducible genomic analysis workflows that can inform environment setup. The Galaxy Training Network offers accessible training on running bioinformatics workflows.

Step 3: Perform Initial Quality Assessment

Run quality assessment on a subset of samples before processing the full dataset. Review the quality reports to identify any systematic issues that require adjustment of preprocessing parameters. This review should include per-base quality, per-sequence quality, GC content, and overrepresented sequences.

Document the quality assessment results for each sample. These records serve as the baseline for evaluating the effectiveness of preprocessing.

Step 4: Configure and Test Preprocessing Parameters

Configure the preprocessing parameters based on the quality assessment results and the analysis goals. Test the parameters on a small number of samples and review the output to verify that the processing achieves the intended results.

Check that adapter contamination is removed, that quality distributions are improved, and that the number of reads retained is reasonable. The proportion of reads retained after preprocessing is an important quality metric that should be recorded for each sample.

Step 5: Process All Samples

Run the preprocessing workflow on all samples using identical parameters. Monitor the processing to identify any samples that fail or produce unexpected results. Samples with very different read retention rates or quality distributions may indicate sample specific issues that require investigation.

Step 6: Verify Preprocessing Results

After preprocessing, verify that the output files are complete and correctly formatted. Check that read counts match expectations, that paired-end files remain properly paired, and that quality metrics have improved as expected.

Generate summary statistics for the cleaned data and compare them with the raw data statistics. This comparison documents the effect of preprocessing and provides quality metrics for reporting.

Step 7: Document and Archive

Document all preprocessing decisions, including tool versions, parameter settings, reference database versions, and quality metrics. This documentation is essential for reproducibility and for interpreting downstream analysis results.

Archive the raw data, the cleaned data, and the documentation according to your institution's data management policies. The NCBI provides guidance on data submission and archival that can inform local practices.

Records and Measurements

Essential Records for Preprocessing

Maintain records of the following for each sample or dataset:

Raw read counts and total bases before preprocessing. These values establish the starting point and allow calculation of retention rates.

Quality metrics from the initial quality assessment, including per-base quality distributions and any identified issues.

Preprocessing parameters, including tool versions, quality thresholds, adapter sequences, and reference database versions.

Cleaned read counts and total bases after preprocessing. The retention rate, calculated as cleaned reads divided by raw reads, is a key quality indicator.

Any samples that failed preprocessing or required special handling, with the reason for the failure or the special handling applied.

Quality Metrics to Track

Per-base quality scores before and after preprocessing. The improvement in quality scores demonstrates the effectiveness of filtering and trimming.

Adapter contamination rates before and after trimming. The residual adapter contamination after trimming should be minimal.

Host contamination rates for shotgun metagenomic samples. The proportion of reads removed as host contamination is informative for understanding sample composition.

Read length distributions before and after trimming. Trimming should produce a predictable length distribution based on the insert size and read length.

GC content distributions before and after preprocessing. Unexpected changes in GC content may indicate the removal of legitimate biological sequences.

Benchmarking and Comparison

For studies comparing multiple samples or conditions, preprocessing metrics should be compared across samples to identify outliers. Samples with substantially different retention rates or quality profiles may require additional investigation.

The comparative analysis of HiSeq3000 and BGISEQ-500 platforms published in Genomics and Informatics demonstrates the value of platform comparison for understanding data characteristics. Similar comparisons within a study can identify batch effects or sample specific issues.

Common Failure Patterns and Troubleshooting

Failure Pattern 1: Overly Aggressive Quality Filtering

Overly aggressive quality filtering removes legitimate biological sequences along with low-quality reads. This pattern is identified by very low read retention rates or by the loss of taxa that are known to be present in the samples.

Troubleshooting involves reviewing the quality score distributions to determine whether the quality threshold is appropriate for the sequencing platform and the downstream analysis. Lowering the quality threshold or using a different filtering strategy may retain more reads without substantially increasing errors.

Failure Pattern 2: Insufficient Adapter Trimming

Insufficient adapter trimming leaves adapter sequences in the cleaned data. These sequences interfere with taxonomic classification and assembly because they do not match any biological reference.

Troubleshooting involves checking the overrepresented sequences in the quality assessment output and verifying that the adapter sequences used for trimming match the adapters used in library preparation. Different library preparation kits use different adapter sequences, and using the wrong adapter sequences will fail to remove contamination.

Failure Pattern 3: Host Filtering Removes Microbial Reads

Host filtering can remove microbial reads that share sequence similarity with the host genome. This pattern is more likely when the host genome contains endogenous viral elements or when microbial sequences have been integrated into the host genome.

The Aedes mariae virome study published in Biology highlights this challenge, noting the potential sequence similarity between viral and mosquito genomes. Troubleshooting involves reviewing the alignment parameters used for host filtering and considering whether the host genome contains sequences that match target organisms.

Failure Pattern 4: Paired-End File Mismatches

Paired-end file mismatches occur when forward and reverse read files contain different numbers of reads or reads in different orders. This pattern causes downstream tools to fail or produce incorrect results.

Troubleshooting involves validating the read files before preprocessing and verifying that the pairing is maintained after each preprocessing step. Some preprocessing tools can reorder reads, and the pairing must be verified after each step.

Failure Pattern 5: Inconsistent Preprocessing Across Samples

Inconsistent preprocessing across samples occurs when parameters are changed between samples or when different tool versions are used. This pattern introduces batch effects that can be misinterpreted as biological variation.

Troubleshooting involves using a workflow management system to ensure consistent processing and documenting any parameter changes. If parameters must be changed, the affected samples should be reprocessed with the final parameters.

Failure Pattern 6: Reference Database Issues

Reference database issues include using outdated databases, using databases with incorrect taxonomy, or using databases that do not cover the target organisms. These issues affect host filtering and downstream taxonomic assignment.

Troubleshooting involves documenting the database version and download date, checking for database updates, and verifying that the database contains the expected reference sequences. The microbial genomics review notes that incomplete and biased reference databases remain a persistent challenge.

Limitations and Interpretation Constraints

Reference Database Limitations

Reference databases are incomplete and biased toward well-studied organisms. Reads from novel or underrepresented taxa may fail to match references and be removed during host filtering or remain unclassified during taxonomic assignment. This limitation is documented in the microbial genomics review published in Infection, Genetics and Evolution.

Researchers should interpret the absence of a taxon in the results as the absence of evidence, not evidence of absence. The proportion of reads that remain unclassified after preprocessing and downstream analysis is an important metric for understanding the limitations of the analysis.

Platform Specific Biases

Different sequencing platforms produce data with different error profiles, quality distributions, and biases. The comparative analysis of HiSeq3000 and BGISEQ-500 platforms published in Genomics and Informatics demonstrates that platform choice affects the characteristics of shotgun metagenomic data.

Preprocessing parameters optimized for one platform may not transfer directly to another platform. Researchers should validate preprocessing parameters for each platform and document platform specific characteristics.

Computational Resource Constraints

Preprocessing large metagenomic datasets requires substantial computational resources, including memory, storage, and processing time. The microbial genomics review notes that computational bottlenecks and economic disparities in sequencing and storage capacities are persistent challenges.

Researchers should plan for computational requirements before beginning analysis and consider using cloud computing or institutional high-performance computing resources when local resources are insufficient.

The Impact of Preprocessing on Downstream Results

Preprocessing decisions directly affect downstream analysis results. Different preprocessing parameters can produce different taxonomic profiles, different assembly results, and different functional annotations from the same raw data.

This sensitivity to preprocessing parameters means that results should be interpreted in the context of the preprocessing decisions made. Comparing results across studies requires understanding the preprocessing approaches used in each study.

Quality Control and Validation

Internal Quality Controls

Internal quality controls include the use of spike-in controls with known sequences, the inclusion of negative controls to identify contamination, and the use of technical replicates to assess reproducibility.

Spike-in controls are known quantities of reference organisms added to samples before sequencing. These controls allow assessment of the entire workflow, including preprocessing, and can identify systematic biases.

Negative controls are samples that contain no biological material but undergo the same library preparation and sequencing. These controls identify reagent contamination and laboratory contamination.

Technical replicates are multiple sequencing runs from the same biological sample. These replicates assess the reproducibility of the entire workflow, including preprocessing.

Validation of Preprocessing Effectiveness

Validation involves checking that preprocessing achieved the intended results. This validation includes reviewing quality metrics before and after preprocessing, checking for residual adapter contamination, and verifying that read pairing is maintained.

For host filtering, validation can include checking that known host sequences are removed and that known microbial sequences are retained. For amplicon data, validation can include checking that primer sequences are removed and that the expected amplicon length distribution is observed.

External Quality Assessment

External quality assessment involves comparing preprocessing results with those from other laboratories or with reference datasets. The EMBL-EBI Training provides resources on bioinformatics data analysis that can inform quality assessment practices.

Participation in community benchmarking efforts can help validate preprocessing approaches and identify areas for improvement. The nf-core community provides standards for pipeline development that include testing and validation components.

Safety and Regulatory Context

Data Management and Privacy

Metagenomic data may contain sequences from human hosts or from pathogens. Researchers must comply with institutional and regulatory requirements for data handling, storage, and sharing.

The NCBI provides guidance on data submission and access that includes considerations for sensitive data. Researchers should review these requirements before beginning analysis and ensure that their data management practices comply.

Responsible Conduct of Research

Reproducible preprocessing is a component of responsible conduct of research. Transparent documentation of preprocessing decisions allows other researchers to evaluate and reproduce the analysis.

The FAIR principles (Findable, Accessible, Interoperable, Reusable) provide a framework for data management that supports reproducibility. The microbial genomics review notes that international initiatives increasingly promote open, interoperable, and reusable bioinformatics infrastructures.

Professional Escalation Criteria

Seek professional guidance or escalate to institutional support when:

Preprocessing results are substantially different from expectations and the cause cannot be identified.

Samples fail preprocessing consistently and the failure indicates a systematic issue with library preparation or sequencing.

The computational requirements exceed available resources and alternative approaches are needed.

Reference database issues cannot be resolved with available expertise.

Results from preprocessing will be used for regulatory submissions or clinical decisions.

Frequently Asked Questions

What is the difference between quality trimming and quality filtering?

Quality trimming removes low-quality bases from the ends of reads while retaining the high-quality portion of the read. Quality filtering removes entire reads that do not meet quality thresholds. Trimming is appropriate when quality declines toward read ends, which is common in Illumina sequencing. Filtering is appropriate when entire reads are low quality, which may indicate failed sequencing reactions. Many preprocessing workflows use both approaches, trimming read ends and then filtering reads that are too short or still too low quality after trimming.

Why is adapter trimming necessary for metagenomic data?

Adapter sequences are added during library preparation and are not part of the biological sequence. When insert sizes are shorter than read lengths, the sequencer reads through the insert into the adapter sequence. These adapter sequences interfere with taxonomic classification because they do not match any biological reference, and they interfere with assembly because they create false overlaps between reads. Adapter trimming removes these sequences before downstream analysis.

How do I choose quality thresholds for preprocessing?

Quality thresholds should be based on the sequencing platform, the downstream analysis requirements, and the quality assessment results. Higher quality thresholds retain fewer reads but produce higher confidence data. Lower quality thresholds retain more reads but include more errors. The optimal threshold balances read retention against error rate for the specific analysis. Review the quality assessment output to understand the quality distribution in your data and test different thresholds on a subset of samples.

What is host contamination and why must it be removed?

Host contamination is DNA from the organism being studied that is present in the sample alongside the microbial community. For example, gut microbiome samples contain host intestinal cells, and insect samples contain insect DNA. Host DNA is not part of the microbial community and must be removed before analysis because it inflates read counts, reduces sensitivity for detecting microbial taxa, and interferes with assembly. Host filtering aligns reads against the host reference genome and removes matching reads.

Should I merge paired-end reads before or after quality filtering?

The order of read merging and quality filtering depends on the analysis goals and the data characteristics. For amplicon data, merging before quality filtering allows quality assessment of the merged product and can improve error correction. For shotgun metagenomic data, reads typically do not overlap and merging is not applicable. The 16S and 23S metabarcoding pipeline described in Biotech performs read merging before quality filtering, which is appropriate for amplicon data with overlapping read pairs.

How do I know if my preprocessing was successful?

Successful preprocessing produces cleaned reads with improved quality metrics, minimal residual adapter contamination, and appropriate read retention rates. Compare quality metrics before and after preprocessing to document the improvement. Check that read counts are reasonable for the sample type and sequencing depth. Verify that paired-end files remain properly paired. The retention rate, calculated as cleaned reads divided by raw reads, should be consistent across samples from the same study.

What records should I keep for reproducible preprocessing?

Keep records of all tool versions, parameter settings, reference database versions and download dates, and quality metrics before and after preprocessing. Document any samples that failed or required special handling. Record the computational environment, including operating system and hardware specifications. This documentation allows the analysis to be repeated or audited and is essential for interpreting downstream results.

When should I seek professional help with preprocessing?

Seek professional help when preprocessing results are consistently unexpected, when samples fail repeatedly, when computational requirements exceed available resources, or when the analysis will be used for regulatory or clinical decisions. Institutional bioinformatics support, collaborators with relevant expertise, and community resources such as the Galaxy Training Network and nf-core documentation can provide guidance.

Related Bioinformatics Guides

• Metagenomics Data Analysis: From Raw Reads to Biological Insights
• RNA-Seq Data Analysis Workflow: From Raw Reads to Insights
• Metabolomics Data Analysis Workflow: From Raw Data to Biological Insight
• Metagenomics Pipeline: From Raw Reads to Taxonomic and Functional Profiles
• Proteomics Data Analysis Workflow: From Raw Spectra to Biological Insights

Related Clinical & Scientific Guides

• A Practical Guide to Detecting Antimicrobial Resistance Genes in Shotgun Metagenomic Data
• Computational Immunology: Modeling the Immune System
• How to Set Hard Filters for Germline Variant Calling: A Practical Guide to GATK Best Practices

References and Further Reading

• NCBI Data Resources. National Center for Biotechnology Information.
• EMBL-EBI Training. European Bioinformatics Institute.
• Bioconductor. Bioconductor Project.
• Galaxy Training Network. Galaxy Project.
• nf-core Documentation. nf-core.
• The Carpentries Lessons. The Carpentries.
• Advancements and challenges in bioinformatics tools for microbial genomics in the last decade: Toward the smart integration of bioinformatics tools, digital resources, and emerging technologies for the analysis of complex biological data.. Infection, genetics and evolution : journal of molecular epidemiology and evolutionary genetics in infectious diseases, 2025.
• IMP: a pipeline for reproducible reference-independent integrated metagenomic and metatranscriptomic analyses.. Genome biology, 2016.
• A workflow for statistical analysis and visualization of microbiome omics data using the R microeco package.. Nature protocols, 2026.
• Navigating prokaryotic viral genome analysis from metagenomic data.. 2026.
• Exploratory Metaviromic Analysis of the Sea-Rock Pool Mosquito &lt,i&gt,Aedes mariae&lt,/i&gt, and the Water of Its Breeding Habitat.. 2026.
• Bioinformatics Strategy for 16s and 23s rRNA Metabarcoding Data.. 2026.
• Comparative analysis of HiSeq3000 and BGISEQ-500 sequencing platform with shotgun metagenomic sequencing data. Genomics and Informatics, 2023.

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