- Multimapping and repetitive regions are important challenges in bioinformatics because they affect the accuracy of read alignment, genome annotation, variant detection, gene expression analysis, and other genomic investigations. During DNA sequencing and RNA sequencing, millions of short or long reads are mapped to a reference genome. Some reads match a single genomic location with high confidence, while others match multiple locations because similar or identical sequences occur in different parts of the genome. These reads are known as multimapping reads. Repetitive regions are genomic sequences that occur more than once, either as identical copies or as related sequences with varying degrees of similarity. Together, multimapping and repetitive DNA can make it difficult to determine the original genomic location of a sequencing read.
- Multimapping occurs when a sequencing read aligns successfully to two or more positions in a reference genome. This may happen because the read originates from a repeated sequence, a duplicated gene, a transposable element, or another region with substantial sequence similarity. If multiple genomic locations produce equally good or nearly equivalent alignments, the aligner may be unable to determine the read’s true origin. Depending on the software and alignment settings, the read may be reported with one primary alignment and several secondary alignments, assigned a low mapping quality, or excluded from certain downstream analyses.
- A simple example illustrates the problem. Suppose a short DNA sequence occurs in three different regions of a reference genome:
Region 1: ATGCTAGCTAGG
Region 2: ATGCTAGCTAGG
Region 3: ATGCTAGCTAGG
- A sequencing read containing this sequence can match all three regions equally well. If the read does not extend into unique surrounding sequences, the aligner cannot determine which genomic copy produced it. The read is therefore multimapping, even though its sequence alignment may be perfect.
- Multimapping is closely related to the structure and organization of eukaryotic genomes. Genomes contain many repeated sequences that have accumulated through duplication, transposition, replication errors, and evolutionary processes. Some repeated sequences are nearly identical, while others have diverged over time. Repetitive DNA may occur in large clusters, dispersed throughout chromosomes, or within genes and regulatory regions. The extent and distribution of repetitive sequences vary among organisms and can represent a substantial proportion of some eukaryotic genomes.
- Repetitive DNA can be classified into several broad categories. Tandem repeats consist of sequence units arranged consecutively, such as satellite DNA, minisatellites, and microsatellites. Interspersed repeats occur at different genomic locations and are often associated with transposable elements. These include long interspersed nuclear elements (LINEs), short interspersed nuclear elements (SINEs), long terminal repeat (LTR) retrotransposons, and DNA transposons. Segmental duplications are relatively large genomic regions that have been copied to other locations and may retain substantial sequence similarity. Duplicated genes and gene families also contain related sequences that can generate ambiguous read alignments.
- Tandem repeats contain DNA motifs repeated one after another. Microsatellites, also called short tandem repeats, generally consist of short repeat units, while minisatellites contain longer repeat units. Satellite DNA may consist of large arrays of repeated sequences, often concentrated in particular chromosomal regions. These sequences are important in chromosome organization, genetic diversity, population studies, and forensic analysis. However, short reads originating from long tandem-repeat arrays may be difficult to assign to a specific repeat copy because many copies share similar sequence patterns.
- Interspersed repetitive elements are distributed across the genome and frequently originate from transposable elements. LINEs and SINEs are examples of non-LTR retrotransposons, while LTR retrotransposons contain long terminal repeat sequences. DNA transposons move through mechanisms involving DNA intermediates. Although many transposable elements are inactive or fragmented, some retain functional activity or influence gene regulation. Reads originating from these elements may map to multiple genomic locations, complicating the analysis of gene expression, genome variation, and regulatory activity.
- Segmental duplications are another important source of multimapping. These regions contain relatively large DNA segments that have been duplicated within a genome and may retain high sequence identity. They can include genes, pseudogenes, regulatory elements, and other genomic features. Because reads from one duplicated copy may closely resemble reads from another, short-read aligners may assign them ambiguously. Segmental duplications are particularly relevant to studies of copy-number variation, gene dosage, inherited disorders, and genome evolution.
- Gene families can also produce multimapping reads. Members of a gene family often share sequence similarity because they evolved through gene duplication followed by divergence. Examples include globin genes, olfactory receptor genes, histone genes, and several immune-related gene families. Reads from highly conserved regions may align to multiple family members, making it difficult to determine which specific gene is expressed or contains a variant. Longer reads or reads spanning unique sequence regions may improve gene-specific assignment.
- Several factors influence the likelihood of multimapping. Read length is particularly important because longer reads are more likely to contain unique sequence information. A short read may fall entirely within a repetitive region, whereas a longer read may extend beyond the repeat into a unique genomic region. Sequence complexity also affects mapping. Low-complexity sequences contain a limited variety of nucleotides or repeated patterns and can match multiple locations. Sequencing errors, genetic variation, reference genome completeness, and the similarity between duplicated regions further influence alignment outcomes.
- Short-read sequencing is particularly vulnerable to multimapping because reads are generally much shorter than many repetitive genomic regions. A read of 50 or 100 bases may be completely contained within a repeat and provide insufficient information for unique mapping. Increasing read length can improve mapping specificity, although it does not solve every problem. Highly similar segmental duplications or nearly identical gene copies may remain difficult to distinguish even with longer reads. Long-read technologies such as Oxford Nanopore and PacBio can help resolve these regions by spanning repeated sequences and extending into unique flanking DNA.
- The alignment software plays an important role in identifying and reporting multimapping reads. Programs such as BWA-MEM, Bowtie2, STAR, HISAT2, and minimap2 use different algorithms and reporting strategies. Some aligners report only the best alignment by default, while others can report multiple candidate alignments under specified settings. In SAM and BAM files, alignments may be classified as primary, secondary, or supplementary. These categories describe different aspects of the alignment representation and should not be confused with the biological certainty of the read’s origin.
- A primary alignment is the alignment record selected as the representative alignment for a read or read segment. Secondary alignments represent alternative mapping locations. Supplementary alignments represent additional alignment segments, often produced when a read aligns in multiple parts, such as when it spans a structural rearrangement or a chimeric sequence. The SAM flag field identifies these record types. Secondary alignments are commonly marked with flag 0x100, while supplementary alignments are marked with flag 0x800. A primary alignment does not necessarily mean that the genomic position is uniquely or confidently determined; it may simply be the aligner’s selected representative record.
- Mapping quality (MAPQ) provides an estimate of confidence in the reported mapping position. Reads with equally good alignments to multiple locations commonly receive low mapping-quality values. However, MAPQ calculation varies among aligners, and some programs use special values or conventions for multimapping reads. A value of 255 in the SAM format indicates that mapping quality is unavailable, not that the mapping is exceptionally reliable. Researchers should therefore consult the documentation of the aligner before interpreting MAPQ values or applying filtering thresholds.
- The NH optional tag is commonly used by some aligners to report the number of reported or identified alignments for a read. The HI tag may identify a particular alignment among multiple alignments. The AS tag commonly stores an alignment score, while the NM tag commonly records edit distance. The availability and precise interpretation of these tags depend on the aligner. They can help identify reads with multiple reported placements, but users should not assume that every aligner writes every tag or uses identical conventions.
- The following command extracts secondary and supplementary alignment records from a BAM file using SAMtools:
samtools view -f 0x100 alignments.bam > secondary_alignments.sam
samtools view -f 0x800 alignments.bam > supplementary_alignments.sam
- The first command selects secondary alignment records, while the second selects supplementary records. These commands do not identify every possible multimapping read in every aligner because some programs may report only a primary placement or use different reporting conventions. A more complete analysis should consider the aligner’s documentation, MAPQ values, optional tags, and alignment-reporting settings.
- Reads with low mapping quality can be filtered using SAMtools. For example:
samtools view -b -q 20 alignments.bam > filtered_alignments.bam
- This command retains alignments with MAPQ values of at least 20. Such filtering may reduce ambiguous mappings, but it can also remove genuine reads from repetitive or duplicated genomic regions. In some analyses, these reads contain important biological information and should not be discarded automatically. The decision to filter them should depend on the research objective and the properties of the reference genome and sequencing data.
- Multimapping has different consequences for different bioinformatics applications. In whole-genome sequencing, ambiguous reads may reduce the accuracy of variant detection in repetitive regions and duplicated genes. A read containing a real variant may be incorrectly assigned to another genomic copy, or a variant may be missed because the supporting reads have low mapping quality. This is particularly important when studying genes with pseudogenes or highly similar paralogs. Specialized methods, longer reads, improved references, and orthogonal validation may be required to distinguish these regions reliably.
- In RNA-seq analysis, multimapping reads can complicate gene and transcript expression quantification. A read derived from a repetitive element may align to many genomic locations, making it difficult to identify the specific source. Similarly, reads from highly similar gene-family members may be assigned ambiguously. If multimapping reads are discarded, expression estimates for some genes or repetitive elements may be underestimated. If they are assigned arbitrarily to one location, expression estimates may become inaccurate. Some quantification programs therefore use probabilistic methods to distribute ambiguous reads among candidate transcripts based on the available evidence.
- Transcript quantification tools such as Salmon and kallisto use lightweight mapping or pseudoalignment strategies to estimate transcript abundance while accounting for ambiguity among similar transcripts. These methods do not necessarily perform conventional full genomic alignment. Instead, they use sequence compatibility and statistical models to estimate the abundance of transcripts that could have generated the observed reads. Alignment-based methods and probabilistic quantification methods therefore address multimapping in different ways, and the choice depends on whether the study requires genomic coordinates, transcript abundance, or detailed splice-junction information.
- Repetitive regions also play an important role in gene expression and epigenetic analysis. Transposable elements may function as promoters, enhancers, transcription-factor binding sites, or other regulatory sequences. Their transcription can be biologically relevant in development, disease, and genome regulation. However, reads derived from repetitive elements may map to multiple copies, making it difficult to quantify their expression at individual genomic locations. Some analyses estimate expression at the family or subfamily level, while others attempt locus-specific quantification using unique sequence information or specialized computational methods.
- In chromatin and epigenomic studies, repetitive regions can also create mapping challenges. ChIP-seq, ATAC-seq, and other assays may generate reads from repetitive DNA, making it difficult to determine the precise genomic origin of a signal. Some pipelines exclude low-mapping-quality reads or mask problematic genomic regions to reduce false-positive signals. However, this can result in loss of information about regulatory activity within repeats. Researchers should therefore distinguish between technical filtering and biological absence of a signal.
- Several bioinformatics tools and databases support the identification and analysis of repetitive regions. RepeatMasker is widely used to identify and annotate interspersed repetitive elements and low-complexity sequences in genomic DNA. It compares sequences with repeat libraries and identifies regions that resemble known repeat families. Dfam provides curated repeat-family models, while Repbase has historically been another important resource for repetitive DNA annotation. Tandem Repeat Finder identifies tandemly repeated sequences, and tools such as RepeatModeler can help construct de novo repeat libraries for genome annotation.
- The following example illustrates how RepeatMasker can be run on a genome sequence: RepeatMasker -species human -pa 8 genome.fasta
- The -species option selects a species-specific repeat library where supported, while -pa specifies the number of parallel processes. The output includes repeat annotations that can be used to identify regions likely to produce ambiguous mappings. For organisms without comprehensive repeat libraries, de novo repeat identification may be needed before genome-wide annotation.
- Low-complexity regions are another important source of ambiguous alignment. These sequences may contain long stretches of a small number of nucleotide types or repeated motifs. Examples include homopolymer runs and simple sequence repeats. Aligners may use masking or filtering methods to reduce the influence of low-complexity regions during mapping. However, masking can remove biologically meaningful information, especially when the analysis focuses on repeat-associated variation or regulatory elements. The choice of masking strategy should therefore be documented.
- Several approaches can reduce or manage multimapping. Increasing read length can provide more unique sequence information and improve the distinction between duplicated regions. Using paired-end reads can also help because the combined information from both ends may identify a unique genomic location even when one end maps to a repeat. Longer insert sizes may provide additional context, although the effectiveness depends on the genomic arrangement and sequencing library.
- Improved reference genomes can also reduce multimapping caused by missing or collapsed genomic regions. A reference assembly that fails to represent duplicated sequences may cause reads from different genomic copies to be assigned to a single region or mapped incorrectly. Long-read sequencing, haplotype-resolved assembly, and improved genome annotation can help represent complex regions more accurately. Pangenome references and variation graphs can provide alternative genomic paths, reducing some forms of bias associated with a single linear reference. However, graph-based mapping does not eliminate all ambiguity, particularly when repeated sequences are identical across multiple paths.
- Specialized mapping methods can also improve analysis in difficult genomic regions. Winnowmap, for example, was developed to improve long-read mapping in repetitive regions through specialized minimizer handling. Graph-based aligners such as GraphAligner can map reads to variation graphs containing alternative genomic sequences. Other specialized tools and analysis strategies may be appropriate for segmental duplications, copy-number variation, and highly repetitive regions. Tool selection should be based on the data type, reference representation, and downstream biological question.
- Long-read sequencing is especially useful for resolving repetitive regions because long reads can span repeat arrays and connect them to unique flanking sequences. PacBio HiFi reads combine long read lengths with high accuracy, while Oxford Nanopore sequencing can generate reads extending across very long genomic regions. These technologies can improve the resolution of structural variants, segmental duplications, tandem repeats, and complex gene clusters. Nevertheless, very long or highly similar repeat arrays may remain difficult to resolve, and assembly quality must be evaluated carefully.
- Multimapping is also relevant to structural-variant analysis. Reads that align in multiple segments or show unusual alignment patterns may provide evidence for insertions, deletions, inversions, duplications, or translocations. However, repetitive regions can generate alignment patterns that resemble structural variants even when no rearrangement exists. Structural-variant callers therefore evaluate split-read evidence, read-pair information, mapping quality, sequence identity, and supporting read counts. Long-read structural-variant tools such as Sniffles and cuteSV use long-read alignment patterns to identify candidate variants, but their results still require quality assessment and careful interpretation.
- The analysis of repetitive regions is important in evolutionary genomics and comparative genomics. Repeats contribute to genome expansion, chromosome organization, gene regulation, and genome evolution. Transposable elements can create new regulatory sequences, contribute to gene duplication, and influence genomic rearrangements. Tandem repeats and microsatellites are also used in population genetics, genetic diversity studies, and forensic identification. Accurate analysis of these regions requires methods that account for repeated sequence copies rather than treating all ambiguous alignments as technical noise.
- For reproducible analysis, researchers should record the reference genome assembly, repeat annotation version, aligner name and version, mapping parameters, multimapping reporting options, MAPQ thresholds, and filtering rules. The handling of secondary and supplementary alignments should also be documented. In expression analysis, researchers should state whether multimapping reads were discarded, probabilistically assigned, or summarized at the repeat-family level. These details are essential because different approaches can produce substantially different results.
- Multimapping and repetitive regions remain major challenges in genome and transcriptome analysis. Repetitive sequences are biologically important components of genomes, but their similarity and repeated distribution make it difficult to assign sequencing reads to unique genomic locations. Understanding the types of repetitive DNA, alignment behavior, mapping-quality conventions, and available computational methods helps researchers interpret ambiguous reads correctly. By combining appropriate sequencing technologies, accurate reference assemblies, specialized aligners, repeat annotation tools, and carefully designed filtering strategies, bioinformaticians can improve the reliability of analyses involving repetitive and duplicated genomic regions.