XSTREME: Comprehensive motif analysis of biological sequence datasets
Abstract
XSTREME is a web-based tool for performing comprehensive motif discovery and analysis in DNA, RNA or protein sequences, as well as in sequences in user-defined alphabets. It is designed for both very large and very small datasets. XSTREME is similar to the MEME-ChIP tool, but expands upon its capabilities in several ways. Like MEME-ChIP, XSTREME performs two types of de novo motif discovery, and also performs motif enrichment analysis of the input sequences using databases of known motifs. Unlike MEME-ChIP, which ranks motifs based on their enrichment in the centers of the input sequences, XSTREME uses enrichment anywhere in the sequences for this purpose. Consequently, XSTREME is more appropriate for motif-based analysis of sequences regardless of how the motifs are distributed within the sequences. XSTREME uses the MEME and STREME algorithms for motif discovery, and the recently developed SEA algorithm for motif enrichment analysis. The interactive HTML output produced by XSTREME includes highly accurate motif significance estimates, plots of the positional distribution of each motif, and histograms of the number of motif matches in each sequences. XSTREME is easy to use via its web server at https://meme-suite.org, and is fully integrated with the widely-used MEME Suite of sequence analysis tools, which can be freely downloaded at the same web site for non-commercial use.
02 /Specification & Classification
Click any tag to browse matching papersIn-depth summary
Claude Opus 5.5 · max effort · 8 Nov 2026 · from the full text of version 1 (3 Sept 2021) · CC BY-NC 4.0
XSTREME couples MEME and STREME de novo discovery with SEA known-motif enrichment, ranking motifs by enrichment anywhere in sequences, unlike center-focused MEME-ChIP; illustrated on Gata1 Cut&Run.
MEME-ChIP ranks motifs by central enrichment, an assumption that can fail for promoter sets, ATAC-seq accessible regions and TF-antibody Cut&Run peaks, where sites may lie off-center or near one end. XSTREME drops positional assumptions, scoring de novo and known motifs on one enrichment scale against a control (shuffled primaries by default) and clustering similar motifs to simplify interpretation.
Key results
ShownOn one Gata1 Cut&Run peak set from erythroid precursor cells, the STREME stage recovered a motif closely matching the known Gata1 binding site, which XSTREME then linked to the corresponding JASPAR Gata1 matrix.
STREME motif 1-SAGATAAGR, shown in the top motif cluster, was associated with JASPAR Gata1 motif MA0035.3; input was the Zhu et al. 2019 peak file GSM4043375 GATA1 D7 S11 peaks.narrowPeak.gz (GEO GSM4043375), using sequences of 400bp to 600bp. · 2 Results; Figure legends
ShownXSTREME's site-position plot revealed a clearly bi-modal distribution of Gata1 sites in this Cut&Run set, consistent with the off-center binding that the authors argue center-based motif ranking handles poorly.
The positional distribution panel of the Fig. 1 output displays clearly bi-modal Gata1 site placement; the authors note that in some Cut&Run experiments most bound sequences have the TF near one end. · 2 Results; Figure legends
ClaimedThe authors argue that ranking by enrichment anywhere in the sequence relative to a control makes XSTREME better suited than MEME-ChIP to promoters, ATAC-seq regions and TF-antibody Cut&Run sets, but no head-to-head comparison is presented.
MEME-ChIP ranks motifs by enrichment in central regions, whereas XSTREME contrasts primary with control sequences (shuffled primaries if none supplied); the abstract also asserts highly accurate significance estimates, yet no MEME-ChIP output, accuracy or calibration metric is reported. · Abstract; 1 Introduction; 2 Results
ShownJoint clustering of de novo and database motifs grouped related signals, placing the known GATA1-TAL1 motif in alignment with the de novo STREME Gata1 motif inside the top reported cluster.
In the Fig. 1 top cluster the GATA1-TAL1 known motif is displayed aligned to STREME motif 1-SAGATAAGR; XSTREME groups discovered and SEA-enriched motifs by mutual similarity and lists groups sorted by statistical significance. · 1 Introduction; 2 Results
ShownPer-sequence match-count histograms expose site multiplicity: in the Gata1 Cut&Run example, about 25% of site-bearing sequences carried more than one predicted motif match.
Histogram in Fig. 1 of predicted motif matches per input sequence, restricted to sequences with at least one match, indicates approximately 25% of site-bearing sequences hold more than one site. · 2 Results
Methods, data and reproducibility
Pipeline for DNA, RNA, protein or custom-alphabet sequences: MEME and STREME discover motifs de novo, SEA scores discovered and database motifs for enrichment in primary versus control sequences (shuffled primaries if none given) on one scale, similar motifs are clustered, and results are reported as interactive HTML.
- Gata1 Cut&Run peaks in erythroid precursor cells from Zhu et al. 2019 (file GSM4043375 GATA1 D7 S11 peaks.narrowPeak.gz), restricted to sequences of 400bp to 600bp · GSM4043375
- Methods & tools
- XSTREME, MEME (multiple EM for motif elicitation), STREME (Sensitive, Thorough, Rapid Enriched Motif Elicitation), SEA (Simple Enrichment Analysis), Motif similarity clustering (algorithm unnamed), MEME-ChIP (discussed comparator; ranks by central enrichment), MEME Suite (web server and non-commercial download), JASPAR transcription factor motif database, CISBP-RNA RNA-binding protein motif database, ELM protein motif database, Cut&Run (antibody-targeted chromatin cleavage assay)
- Reproducibility gaps
- Local MEME Suite download is free for non-commercial use; XSTREME version and run parameters for the Gata1 example are not reported; the analyzed peaks are given only as GEO accession GSM4043375 (Zhu et al. 2019), with no data URL.
Limitations (3)
- The authors note that although XSTREME is offered through a public web server, downloading the MEME Suite for local installation is free only for non-commercial use. · stated by the authors
- Evaluation consists of one illustrative Gata1 Cut&Run run shown as an output screenshot; no benchmark of discovery accuracy, significance calibration or runtime and no side-by-side MEME-ChIP run are reported, so claimed advantages remain untested here. · noted in this summary
- Stated support for RNA, protein and custom alphabets and for very large or very small datasets is not exercised with examples or scaling data, and the shuffling scheme used to build default control sets is not described. · noted in this summary
Who should read:Read this if you need motif discovery plus known-motif enrichment for DNA, RNA or protein sets whose sites are not centered, such as promoters, ATAC-seq regions or Cut&Run peaks, as an alternative to MEME-ChIP.
Reproducibility ledger
Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.
Sign in to file a reproducibility report.
Related papers
- STREME: Accurate and versatile sequence motif discovery
Timothy L. Bailey·BioinformaticsPeer-reviewed·Mar 23, 2021
- SEA: Simple Enrichment Analysis of motifs
Timothy L. Bailey, Charles E. Grant·bioRxiv·Aug 23, 2021
- Regmex, Motif analysis in ranked lists of sequences
Morten Muhlig Nielsen, Paula Tataru, … Jakob Skou Pedersen·Algorithms for Molecular BiologyPeer-reviewed·Dec 8, 2018
- Factorbook Motif Pipeline: A de novo motif discovery and filtering web server for ChIP-seq peaks
Bong-Hyun Kim, Jiali Zhuang, … Zhiping Weng·bioRxiv·Dec 4, 2015
- Memes: an R interface to the MEME Suite
Spencer L. Nystrom, Daniel J. McKay·bioRxiv·Apr 23, 2021
- MotiMul: A significant discriminative sequence motif discovery algorithm with multiple testing correction
Koichi Mori, Haruka Ozaki, Tsukasa Fukunaga·IEEE International Conference on Bioinformatics and Biomedicine (BIBM)Peer-reviewed·Dec 16, 2020