[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7EC5M7XM2WT4RMD5JRFANXB":3,"skill-same:01M7EC5M7XM2WT4RMD5JRFANXB":62},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":30,"commitSha":31,"skillDirectory":32,"forksCount":14,"repositoryHtmlUrl":33,"repositoryHomepageUrl":33,"ownerHtmlUrl":34,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":35,"install":36,"resources":49},"01M7EC5M7XM2WT4RMD5JRFANXB","bio-clip-seq-differential-clip",null,"Identify differentially bound regions across CLIP-seq conditions (knockdown vs control, treatment vs vehicle, disease vs healthy) using DEWSeq (sliding-window DESeq2), Flipper (Skipper-downstream), ASpeak, edgeR, or limma-voom. Use when computing condition-level changes in RBP binding intensity, choosing peak-level vs window-level vs crosslink-level testing, designing replicate experiments, or distinguishing biological binding shifts from technical confounders.","Mr-Nobody-Anonymous\u002FAll-skills","Mr-Nobody-Anonymous","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F184253119?v=4","01M7AEXFJHKTYQW5R0ZCQQTWFV","anonymous","01M7C00M3GQR4QKP9GZA42GFAK",0,"MIT","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fbiology\u002Fbio-clip-seq-differential-clip\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: bio-clip-seq-differential-clip\ndescription: Identify differentially bound regions across CLIP-seq conditions (knockdown vs control, treatment vs vehicle, disease vs healthy) using DEWSeq (sliding-window DESeq2), Flipper (Skipper-downstream), ASpeak, edgeR, or limma-voom. Use when computing condition-level changes in RBP binding intensity, choosing peak-level vs window-level vs crosslink-level testing, designing replicate experiments, or distinguishing biological binding shifts from technical confounders.\ntool_type: r\nprimary_tool: DEWSeq\nsource: \"https:\u002F\u002Fgithub.com\u002FGPTomics\u002FbioSkills\"\nsource_repository: \"GPTomics\u002FbioSkills\"\nsource_path: \"clip-seq\u002Fdifferential-clip\u002FSKILL.md\"\nlicense: \"MIT\"\nimported_at: \"2026-09-20\"\n---\n\n## Version Compatibility\n\nReference examples tested with: DEWSeq 1.18+, htseq-clip 2.0+, DESeq2 1.44+, edgeR 4.2+, limma 3.60+, Flipper (commit 2024.04+), Skipper (commit 2023.05+), pybedtools 0.10+, pyranges 0.0.129+.\n\nBefore using code patterns, verify installed versions match. If versions differ:\n- R: `packageVersion('\u003Cpkg>')` then `?function_name` to verify parameters\n- Python: `pip show \u003Cpackage>` then `help(module.function)` to check signatures\n- CLI: `\u003Ctool> --version` then `\u003Ctool> --help` to confirm flags\n\nIf code throws unexpected errors, introspect the installed package and adapt the example to match the actual API rather than retrying.\n\n# Differential CLIP-seq Analysis\n\n**\"Identify regions with changed RBP binding across conditions\"** -> Test for condition-level differences in IP enrichment relative to SMInput, accounting for replicate variance and (where available) sequencing depth normalization. Three statistical scales are possible: peak-level (test each peak as a unit), window-level (test fixed transcriptome windows; DEWSeq, Flipper), or crosslink-site level (test single-nt positions). The choice depends on the biology (narrow regulatory shift vs broad binding-mode change) and on which upstream peak caller was used (CLIPper -> peak-level; Skipper -> window-level; PureCLIP -> CL-level).\n\n- R (window-level, DEWSeq + htseq-clip): `library(DEWSeq); dds \u003C- DESeqDataSetFromSlidingWindows(counts, colData, design=~condition); dds \u003C- DESeq(dds); res \u003C- results(dds)`\n- R (Skipper downstream, Flipper): `flipper differential -i skipper_out\u002F --design design.tsv --contrast treatment vs control -o flipper_out\u002F`\n- R (peak-level, edgeR): `dge \u003C- DGEList(counts=peak_counts, group=condition); dge \u003C- calcNormFactors(dge); design \u003C- model.matrix(~condition); dge \u003C- estimateDisp(dge, design); fit \u003C- glmQLFit(dge, design); res \u003C- glmQLFTest(fit)`\n- R (peak-level, limma-voom): `v \u003C- voom(dge, design); fit \u003C- lmFit(v, design); fit \u003C- eBayes(fit); res \u003C- topTable(fit, coef=2, number=Inf)`\n- CLI (htseq-clip preprocessing): `htseq-clip extract -i annotation.gff -o annotation_windows.bed -w 50 -s 20 && htseq-clip count -i sample.bam -w annotation_windows.bed -o sample.counts.txt`\n\nDEWSeq is the EMBL\u002FHentze-group windowed-NB approach for CLIP binding-site discovery, commonly adapted for differential testing via the interaction design. Flipper is the Skipper-companion tool (Flanagan 2026) for the modern Skipper workflow. Peak-level edgeR\u002Flimma-voom work when the upstream peak caller produced a comparable peak BED across conditions (e.g., consensus peaks from CLIPper).\n\n## Algorithmic Taxonomy\n\n| Tool | Scale | Statistical model | Replicate requirement | Strength | Fails when |\n|------|-------|-------------------|----------------------|----------|------------|\n| DEWSeq (Schwarzl 2024) | 50-100 nt sliding window | Negative binomial GLM (DESeq2 internals) | >= 2 reps per condition | Designed specifically for CLIP; integrates SMInput | Slow on dense libraries; output window-resolution |\n| Flipper (Flanagan 2026) | Skipper window (100 nt) | Negative binomial; designed for Skipper output | >= 2 reps per condition | Modern; pairs with Skipper peak caller | Only useful if upstream is Skipper |\n| ASpeak | Peak | Negative binomial | >= 2 reps | Peak-level for CLIPper output | Less popular; legacy |\n| edgeR (general) | Peak or window | Quasi-likelihood F-test on NB | >= 2 reps | Mature, widely cited | Generic; not CLIP-aware; needs careful normalization |\n| limma-voom | Peak or window | Linear model with mean-variance trend | >= 2 reps | Fast; well-validated; handles small samples | Generic; treats counts as continuous after voom |\n| DESeq2 (direct) | Peak or window | Negative binomial GLM | >= 2 reps | Mature; same engine as DEWSeq | Same as edgeR caveats |\n| Single-cell CLIP (specialized) | Cell-resolved (scCLIP) | Cell-mixture \u002F pseudobulk | Cells | Single-cell CLIP differential | Nascent; few published tools |\n| diffbind (CLIP adaptation) | Peak | DESeq2 or edgeR backend | >= 2 reps | Familiar from ChIP\u002FATAC | Designed for ChIP; needs CLIP-specific normalization |\n| MAnorm2 | Peak | Hierarchical empirical Bayes | >= 2 reps | Tested on ChIP; less on CLIP | Less CLIP-specific |\n\nMethodology evolves; verify DEWSeq vignette and Flipper paper for current best practice. The DEWSeq + htseq-clip pipeline (Schwarzl 2024) is the most-published CLIP-specific differential framework; Flipper (Flanagan 2026) is the modern Skipper-coupled alternative.\n\n## Critical Decision: The Interaction-Term Design\n\n**Use `~ type + condition + type:condition` and test the interaction-term coefficient.** This is the single most consequential statistical choice in differential CLIP.\n\n- `type` = `ip` vs `sminput` (whether the library is IP or size-matched input)\n- `condition` = `treated` vs `untreated` (or KD vs control)\n- `type:condition` interaction = \"Does the IP-vs-input ratio shift across conditions?\"\n\nA naive `~ condition` design tests whether read counts differ regardless of whether they come from IP or SMInput - this confounds binding changes with expression changes. The interaction term explicitly tests for **differential binding** (the biologically meaningful signal) rather than differential expression at peak loci.\n\nWhen testing, extract the interaction-term coefficient: `results(dds, name = 'typeip.conditiontreat')`. The log2FoldChange returned is the change in IP\u002Finput ratio in `treated` vs `untreated` - this is what \"differential binding\" means.\n\n## Critical Choice: Peak-Level vs Window-Level vs Crosslink-Level\n\nThree scales differ in resolution and statistical power:\n\n**Peak-level (edgeR, limma-voom, DESeq2 on CLIPper peaks):** Test each consensus peak's IP\u002FSMInput log2 FC across conditions. Pro: peak boundaries are biologically meaningful; output interpretable. Con: peak set changes across conditions (a new peak in treatment but missing in control complicates testing); SMInput normalization must be applied consistently.\n\n**Window-level (DEWSeq, Flipper):** Tile transcriptome into fixed 50-100 nt windows; test each window. Pro: comparable across conditions (windows are pre-defined); handles binding-mode shifts within a peak; high statistical power. Con: window-resolution; multiple-testing burden (millions of windows); biological meaning of a window needs translation.\n\n**Crosslink-level (custom):** Test each single-nt CL position. Pro: nucleotide resolution; captures motif-level shifts. Con: very low coverage per position; massive multiple-testing burden; rarely used in published differential CLIP.\n\n| Goal | Scale | Tool |\n|------|-------|------|\n| ENCODE-style peak-level differential (CLIPper upstream) | Peak | DESeq2 \u002F edgeR \u002F limma-voom on CLIPper consensus peaks |\n| Maximum sensitivity windowed differential | Window | DEWSeq (with htseq-clip) or Flipper (with Skipper) |\n| Modern Skipper-coupled workflow | Window | Flipper |\n| Single-cell scCLIP differential | Cell | Specialized single-cell CLIP methods (few published tools) |\n| Compare binding-mode shifts within a peak | Window | DEWSeq |\n| Allele-specific differential | CL site | BEAPR + custom logistic |\n| RBP-KD effect on binding profile | Peak\u002Fwindow | DEWSeq (handles KD-effect on RBP itself) |\n\n## DEWSeq Workflow (Window-Level Differential)\n\nDEWSeq is the EMBL\u002FHentze-group windowed-NB framework for CLIP binding-site discovery, adapted here for differential testing via the interaction design. The pipeline is:\n\n**Goal:** Identify transcriptome windows where the IP-vs-SMInput ratio shifts across conditions, accounting for replicate variance with the negative-binomial GLM.\n\n**Approach:** Use htseq-clip to extract sliding 50 nt windows across annotated features, count reads per window per sample, build a DESeqDataSetFromSlidingWindows object with the `~ type + condition + type:condition` interaction design, extract the `typeip.conditiontreat` interaction coefficient as the differential-binding effect size, and aggregate adjacent significant windows with `bedtools merge -d 100`.\n\n```bash\n# Step 1: htseq-clip generates sliding-window count matrices\nhtseq-clip extract \\\n    -i gencode.v38.annotation.gff \\\n    -o annotation_windows.bed \\\n    --window-size 50 \\\n    --window-step 20 \\\n    --feature-type CDS,UTR\n\n# Step 2: count IP and SMInput reads per window per sample\nfor sample in ip_rep1 ip_rep2 sminput_rep1 sminput_rep2; do\n    htseq-clip count \\\n        -i ${sample}.dedup.bam \\\n        -a annotation_windows.bed \\\n        -o ${sample}.counts.txt \\\n        --mate 2\ndone\n# (For eCLIP, --mate 2 because R2 5' is the truncation site; for iCLIP single-end use --mate 1)\n\n# Step 3: DEWSeq differential testing\nhtseq-clip mergeCounts \\\n    -i ip_rep1.counts.txt ip_rep2.counts.txt sminput_rep1.counts.txt sminput_rep2.counts.txt \\\n    -o merged_counts.tsv\n```\n\n```r\nlibrary(DEWSeq)\n\ncounts \u003C- read.table('merged_counts.tsv', sep='\\t', header=TRUE, row.names=1)\ncolData \u003C- data.frame(\n    sample = c('ip_rep1','ip_rep2','sminput_rep1','sminput_rep2'),\n    type = c('ip','ip','sminput','sminput'),\n    condition = c('treated','treated','untreated','untreated')\n)\n\ndds \u003C- DESeqDataSetFromSlidingWindows(\n    countData = counts,\n    colData = colData,\n    annotObj = 'annotation_windows.bed',\n    design = ~ type + condition\n)\n\ndds \u003C- DESeq(dds)\n# In a model with `~ type + condition`, the IP-vs-input contrast is the simple condition\n# main effect; to test the differential CLIP signal between conditions use the interaction\n# term name from the design matrix (matches the skill's interaction-term guidance below):\nres \u003C- results(dds, name='typeip.conditiontreated')\n\n# Window-level FDR adjustment\nres_filtered \u003C- res[!is.na(res$padj) & res$padj \u003C 0.05 & abs(res$log2FoldChange) > 1, ]\n\n# Aggregate adjacent significant windows into differential regions\nsig_windows \u003C- as.data.frame(res_filtered)\nsig_windows$chr \u003C- gsub('_.*', '', rownames(sig_windows))\n# Custom reduce: combine adjacent windows within 100 nt\n```\n\n## Flipper Workflow (Skipper-Coupled)\n\nFlipper (Flanagan 2026) is the differential companion to Skipper, operating on the same 100 nt feature-respecting windows with a negative-binomial (DESeq2-based) differential test.\n\n```bash\n# Assume Skipper has been run on all samples; Skipper output is at skipper_out\u002F\nflipper differential \\\n    -i skipper_out\u002F \\\n    --design design.tsv \\\n    --contrast treatment vs control \\\n    -o flipper_out\u002F\n\n# design.tsv format:\n# sample_id   condition   replicate   ip_or_input\n# ip_treat_r1 treatment   1           ip\n# ip_treat_r2 treatment   2           ip\n# in_treat_r1 treatment   1           input\n# ...\n```\n\nOutput: differential window BED with log2 FC, p, padj per window.\n\n## Peak-Level Differential (CLIPper Upstream)\n\n```r\nlibrary(DESeq2)\nlibrary(GenomicRanges)\nlibrary(Rsubread)\n\n# Step 1: union of CLIPper peaks across conditions\n# (See bedtools merge upstream)\npeaks \u003C- read.table('consensus_peaks.bed', sep='\\t', col.names=c('chr','start','end','name','score','strand'))\n\n# Step 2: count reads per peak per sample with featureCounts\nsaf \u003C- data.frame(\n    GeneID = peaks$name,\n    Chr = peaks$chr,\n    Start = peaks$start + 1,  # 1-based for featureCounts\n    End = peaks$end,\n    Strand = peaks$strand\n)\ncounts_ip \u003C- featureCounts(\n    files = c('ip_treat_r1.bam','ip_treat_r2.bam','ip_ctrl_r1.bam','ip_ctrl_r2.bam'),\n    annot.ext = saf,\n    isGTFAnnotationFile = FALSE,\n    strandSpecific = 1,\n    isPairedEnd = TRUE\n)$counts\n\ncounts_in \u003C- featureCounts(\n    files = c('in_treat_r1.bam','in_treat_r2.bam','in_ctrl_r1.bam','in_ctrl_r2.bam'),\n    annot.ext = saf,\n    isGTFAnnotationFile = FALSE,\n    strandSpecific = 1,\n    isPairedEnd = TRUE\n)$counts\n\n# Step 3: DESeq2 with IP vs SMInput interaction\nall_counts \u003C- cbind(counts_ip, counts_in)\ncolData \u003C- data.frame(\n    type = rep(c('ip','input'), each=4),\n    condition = rep(c('treat','treat','ctrl','ctrl'), 2),\n    replicate = rep(c('r1','r2','r1','r2'), 2)\n)\n\ndds \u003C- DESeqDataSetFromMatrix(countData = all_counts, colData = colData,\n                               design = ~ type + condition + type:condition)\ndds \u003C- DESeq(dds)\n\n# The interaction term `typeip.conditiontreat` tests:\n# does the IP\u002Finput ratio differ in treatment vs control?\nres \u003C- results(dds, name = 'typeip.conditiontreat')\n\n# Filter\nres_sig \u003C- res[!is.na(res$padj) & res$padj \u003C 0.05 & abs(res$log2FoldChange) > 1, ]\n```\n\nThe interaction-term design (`type:condition`) is the correct statistical model for differential CLIP: it tests whether the IP-vs-input ratio differs across conditions, which is what \"differential binding\" means. Naive testing of just `condition` (ignoring SMInput) confounds binding changes with expression changes.\n\n## RBP Knockdown Experiment Design\n\nThe canonical differential CLIP design is to knock down the RBP and observe what binding sites are lost. Caveats:\n\n| Issue | Implication | Mitigation |\n|-------|-------------|------------|\n| RBP KD also depletes the RBP protein in cells | siRNA\u002FshRNA reduces RBP -> reduces IP yield -> reduces unique fragments | Normalize against SMInput WITHIN each condition; the relative IP\u002FSMInput captures binding, not protein level |\n| RBP KD changes transcript abundance | mRNA stability regulators (HuR, PUM2) when knocked down change target abundance | Both IP and SMInput see the change; ratio still works |\n| Off-target effects of siRNA | Multiple binding profiles change | Use multiple independent siRNAs; require concordance |\n| KD efficiency varies | Lower KD -> smaller binding-loss signal | Validate KD by WB on the same IP lysate; > 70% protein loss target |\n| Rescue requires re-introducing RBP | siRNA-resistant RBP cDNA for rescue | The standard differential validation experiment |\n\n## Per-Tool Failure Modes\n\n### DEWSeq -- Slow on dense libraries\n\n**Trigger:** Whole-genome window tiling at 20 nt step; dense library (50M unique fragments); 4+ samples.\n\n**Mechanism:** DEWSeq runs DESeq2 internals on millions of windows; the dispersion fit on this many features is slow.\n\n**Symptom:** Runtime > 6 h; out-of-memory; \"size of object exceeds vector limit\".\n\n**Fix:** Increase window step size to 50 nt; pre-filter windows with low counts; or restrict to expressed transcripts only. DEWSeq vignette suggests `keep \u003C- rowSums(counts(dds)) >= 30; dds \u003C- dds[keep,]` before testing.\n\n### DEWSeq -- Custom adjacency aggregation needed\n\n**Trigger:** Windows are 50 nt; biological binding sites are 50-500 nt; user expects DEWSeq to output continuous \"differential regions\" but gets individual windows.\n\n**Mechanism:** DEWSeq outputs per-window results; aggregating adjacent significant windows into regions is a separate step.\n\n**Symptom:** Output has 10,000 individual windows; user expects 1,000 biological regions.\n\n**Fix:** Use the DEWSeq utility `resultsDEWSeq()` then `bedtools merge -d 100` on the significant-window BED. Or use the `top_hits_to_bed.R` script from DEWSeq examples.\n\n### Peak-level differential -- Peak set differs between conditions\n\n**Trigger:** CLIPper called peaks separately per condition; treatment has peaks at sites missing in control (and vice versa).\n\n**Mechanism:** Peak unification requires a consensus peakset; testing on a \"treatment-only\" peak underestimates evidence in control (zero reads) and produces spurious DE.\n\n**Symptom:** \"Treatment-specific\" peaks dominate DE results; biologically implausible.\n\n**Fix:** Generate consensus peakset across all conditions (bedtools merge of all per-condition peak BEDs); count reads per consensus peak across all samples; THEN run differential. The Yeo lab convention is consensus peakset across all samples.\n\n### Interaction term forgotten\n\n**Trigger:** DESeq2 design `~ condition` instead of `~ type + condition + type:condition`.\n\n**Mechanism:** Simple `~ condition` tests whether read counts differ between treatment and control regardless of whether reads are from IP or SMInput. A condition-driven expression change in SMInput is detected as DE binding.\n\n**Symptom:** DE results dominated by transcripts with global expression changes (housekeeping shifts).\n\n**Fix:** Always use the interaction-term design. The biologically meaningful test is the interaction `type:condition` p-value.\n\n### Normalization assumptions\n\n**Trigger:** edgeR `calcNormFactors(method='TMM')` on CLIP-seq data.\n\n**Mechanism:** TMM assumes most features (genes) are not differentially expressed. CLIP-seq peak counts can be globally shifted if the RBP itself is knocked down; TMM normalization would force the shift to be invisible.\n\n**Symptom:** Knockdown experiment shows ~0 DE peaks; expected hundreds.\n\n**Fix:** Use SMInput as the control library; spike-in normalization if available; or skip TMM and use library-size normalization only. Some CLIP-specific tools (DEWSeq, Flipper) handle this internally.\n\n### Flipper requires Skipper upstream\n\n**Trigger:** Flipper called on CLIPper output.\n\n**Mechanism:** Flipper expects Skipper's window-level output format with beta-binomial estimates.\n\n**Symptom:** Flipper crashes or produces nonsense.\n\n**Fix:** Use DEWSeq with htseq-clip for CLIPper-upstream workflows; use Flipper only with Skipper.\n\n## Decision Tree by Scenario\n\n| Scenario | Tool + design | Why |\n|----------|---------------|-----|\n| KD vs control eCLIP, CLIPper upstream | DEWSeq + htseq-clip | CLIP-specific NB GLM with interaction term |\n| KD vs control eCLIP, Skipper upstream | Flipper | Skipper companion; matches windowing |\n| Treatment vs vehicle (small effect) | DEWSeq (window-level higher power) | Sliding windows capture small shifts |\n| Multiple time points | DEWSeq with time as covariate | Continuous design with time vector |\n| Allele-specific differential | BEAPR per-allele + custom logistic | See clip-seq\u002Fclip-alignment for WASP |\n| Single-cell CLIP differential | Specialized single-cell CLIP methods | Nascent; few published tools |\n| Differential motif occupancy | Window-level + DEWSeq + motif overlap | Combine differential windows with motif BED |\n| RBP overexpression vs control | Same as KD reversed | Same statistical framework |\n| Compare two RBPs | NOT differential CLIP; use SPIDR or separate CLIPs | Different RBPs need separate IPs |\n| Spike-in normalization needed | DEWSeq + spike-in size factors | For global occupancy shifts |\n| chimeric eCLIP differential miRNA targets | Custom; treat each miRNA-target chimera as feature | Specialized; see clip-seq\u002Fago-clip-mirna-targets |\n\n## Reconciliation: When Differential Tools Disagree\n\n| Pattern | Likely cause | Action |\n|---------|--------------|--------|\n| DEWSeq finds many DE windows; edgeR peak-level finds few | Window-level higher power for narrow shifts | Aggregate DEWSeq windows; cross-check |\n| edgeR many DE; DEWSeq few | edgeR not accounting for SMInput | Re-run edgeR with interaction term |\n| DE peaks dominated by expression changes | No interaction term | Use `~ type + condition + type:condition` |\n| KD experiment shows ~0 DE | TMM over-corrects global shift | Switch to library-size norm only; use SMInput |\n| siRNA replicates discordant | Off-target effects vary | Use multiple independent siRNAs; require concordance |\n| Treatment-only peaks dominate DE | No consensus peakset | Generate consensus first; then test on unified set |\n| Significant windows scattered | Window aggregation step skipped | `bedtools merge -d 100` on significant-window BED |\n| Same gene appears in many DE windows | Multiple binding sites per gene differential | Report at gene-level too; not just window-level |\n| Flipper fails with non-Skipper input | Upstream mismatch | Use DEWSeq for non-Skipper workflows |\n| DESeq2 dispersion fit fails | Too few replicates (n=2 per condition); too few features after filtering | Increase replicates; or relax filtering |\n\n**Operational rule for high-confidence differential reporting:** (a) Use SMInput-aware design (`~ type + condition + type:condition`); (b) generate consensus peakset across conditions; (c) require padj \u003C 0.05 AND |log2FC| > 1; (d) require concordance with at least one orthogonal method (e.g., DEWSeq + edgeR peak-level on same data); (e) for KD experiments, validate KD efficiency by WB and require multiple independent siRNAs.\n\n## Common Errors\n\n| Error \u002F symptom | Cause | Solution |\n|-----------------|-------|----------|\n| DESeq2 `~ condition` finds many DE | No interaction with type | Use `~ type + condition + type:condition` and test interaction |\n| DEWSeq output is gene-level not window-level | `keep \u003C- ...` filter too aggressive | Loosen pre-filter |\n| Adjacent significant windows not aggregated | Forgot bedtools merge | `bedtools merge -d 100` on sig windows |\n| edgeR TMM fits all libraries to one value | Most-features-not-DE assumption violated | Use library-size norm; or DEWSeq for CLIP-specific |\n| Flipper crashes on CLIPper input | Tool mismatch | Switch to DEWSeq |\n| Few replicates -> unstable estimates | n=2 not enough for dispersion | Increase n; or use limma-voom (more tolerant) |\n| Peaks differ across conditions | Per-condition peak calls | Unify with consensus peakset |\n| KD experiment yields 0 DE | Normalization over-corrected; OR KD efficiency too low | Validate KD WB; check normalization |\n| Global shift in IP relative to SMInput | RBP itself knocked down so IP yield lower | Normalize WITHIN each condition |\n| Lots of \"treatment-only\" peaks | Caller stringency higher in one condition | Use consensus peakset for fairness |\n\n## References\n\n- Schwarzl T et al 2024 Nucleic Acids Res 52:e1 (DEWSeq, windowed NB binding-site discovery; DESeq2-based, adaptable to differential designs)\n- Sahadevan S et al 2022 Methods Mol Biol 2404:189 (DEWSeq + htseq-clip pipeline)\n- Flanagan K, Xu S, Yeo GW 2026 bioRxiv 2026.03.13.711628 (Flipper, Skipper-companion differential; preprint)\n- Boyle EA et al 2023 Cell Genomics 3:100317 (Skipper, parent of Flipper)\n- Love MI et al 2014 Genome Biol 15:550 (DESeq2)\n- McCarthy DJ et al 2012 Nucleic Acids Res 40:4288 (edgeR)\n- Ritchie ME et al 2015 Nucleic Acids Res 43:e47 (limma)\n- Yang EW et al 2019 Nat Commun 10:1338 (BEAPR allele-specific protein-RNA binding)\n- Van Nostrand EL et al 2020 Nature 583:711 (ENCODE 150 RBP shRNA + eCLIP comparison)\n\n## Related Skills\n\n- clip-seq\u002Fclip-peak-calling - CLIPper \u002F Skipper outputs feed differential\n- clip-seq\u002Fclip-qc - Replicate QC required for valid differential\n- clip-seq\u002Fbinding-site-annotation - Annotate differential regions\n- clip-seq\u002Fclip-motif-analysis - Motif analysis on differential windows\n- clip-seq\u002Fago-clip-mirna-targets - Differential miRNA targeting from chimeric eCLIP\n- differential-expression\u002Fdeseq2-basics - Underlying NB GLM model\n- differential-expression\u002Fde-results - DE results interpretation\n- differential-expression\u002Fedger-basics - edgeR for peak counts\n- chip-seq\u002Fdifferential-binding - DNA-protein analogue\n","## Version Compatibility\n\nReference examples tested with: DEWSeq 1.18+, htseq-clip 2.0+, DESeq2 1.44+, edgeR 4.2+, limma 3.60+, Flipper (commit 2024.04+), Skipper (commit 2023.05+), pybedtools 0.10+, pyranges 0.0.129+.\n\nBefore using code patterns, verify installed versions match. If versions differ:\n- R: `packageVersion('\u003Cpkg>')` then `?function_name` to verify parameters\n- Python: `pip show \u003Cpackage>` then `help(module.function)` to check signatures\n- CLI: `\u003Ctool> --version` then `\u003Ctool> --help` to confirm flags\n\nIf code throws unexpected errors, introspect the installed package and adapt the example to match the actual API rather than retrying.\n\n# Differential CLIP-seq Analysis\n\n**\"Identify regions with changed RBP binding across conditions\"** -> Test for condition-level differences in IP enrichment relative to SMInput, accounting for replicate variance and (where available) sequencing depth normalization. Three statistical scales are possible: peak-level (test each peak as a unit), window-level (test fixed transcriptome windows; DEWSeq, Flipper), or crosslink-site level (test single-nt positions). The choice depends on the biology (narrow regulatory shift vs broad binding-mode change) and on which upstream peak caller was used (CLIPper -> peak-level; Skipper -> window-level; PureCLIP -> CL-level).\n\n- R (window-level, DEWSeq + htseq-clip): `library(DEWSeq); dds \u003C- DESeqDataSetFromSlidingWindows(counts, colData, design=~condition); dds \u003C- DESeq(dds); res \u003C- results(dds)`\n- R (Skipper downstream, Flipper): `flipper differential -i skipper_out\u002F --design design.tsv --contrast treatment vs control -o flipper_out\u002F`\n- R (peak-level, edgeR): `dge \u003C- DGEList(counts=peak_counts, group=condition); dge \u003C- calcNormFactors(dge); design \u003C- model.matrix(~condition); dge \u003C- estimateDisp(dge, design); fit \u003C- glmQLFit(dge, design); res \u003C- glmQLFTest(fit)`\n- R (peak-level, limma-voom): `v \u003C- voom(dge, design); fit \u003C- lmFit(v, design); fit \u003C- eBayes(fit); res \u003C- topTable(fit, coef=2, number=Inf)`\n- CLI (htseq-clip preprocessing): `htseq-clip extract -i annotation.gff -o annotation_windows.bed -w 50 -s 20 && htseq-clip count -i sample.bam -w annotation_windows.bed -o sample.counts.txt`\n\nDEWSeq is the EMBL\u002FHentze-group windowed-NB approach for CLIP binding-site discovery, commonly adapted for differential testing via the interaction design. Flipper is the Skipper-companion tool (Flanagan 2026) for the modern Skipper workflow. Peak-level edgeR\u002Flimma-voom work when the upstream peak caller produced a comparable peak BED across conditions (e.g., consensus peaks from CLIPper).\n\n## Algorithmic Taxonomy\n\n| Tool | Scale | Statistical model | Replicate requirement | Strength | Fails when |\n|------|-------|-------------------|----------------------|----------|------------|\n| DEWSeq (Schwarzl 2024) | 50-100 nt sliding window | Negative binomial GLM (DESeq2 internals) | >= 2 reps per condition | Designed specifically for CLIP; integrates SMInput | Slow on dense libraries; output window-resolution |\n| Flipper (Flanagan 2026) | Skipper window (100 nt) | Negative binomial; designed for Skipper output | >= 2 reps per condition | Modern; pairs with Skipper peak caller | Only useful if upstream is Skipper |\n| ASpeak | Peak | Negative binomial | >= 2 reps | Peak-level for CLIPper output | Less popular; legacy |\n| edgeR (general) | Peak or window | Quasi-likelihood F-test on NB | >= 2 reps | Mature, widely cited | Generic; not CLIP-aware; needs careful normalization |\n| limma-voom | Peak or window | Linear model with mean-variance trend | >= 2 reps | Fast; well-validated; handles small samples | Generic; treats counts as continuous after voom |\n| DESeq2 (direct) | Peak or window | Negative binomial GLM | >= 2 reps | Mature; same engine as DEWSeq | Same as edgeR caveats |\n| Single-cell CLIP (specialized) | Cell-resolved (scCLIP) | Cell-mixture \u002F pseudobulk | Cells | Single-cell CLIP differential | Nascent; few published tools |\n| diffbind (CLIP adaptation) | Peak | DESeq2 or edgeR backend | >= 2 reps | Familiar from ChIP\u002FATAC | Designed for ChIP; needs CLIP-specific normalization |\n| MAnorm2 | Peak | Hierarchical empirical Bayes | >= 2 reps | Tested on ChIP; less on CLIP | Less CLIP-specific |\n\nMethodology evolves; verify DEWSeq vignette and Flipper paper for current best practice. The DEWSeq + htseq-clip pipeline (Schwarzl 2024) is the most-published CLIP-specific differential framework; Flipper (Flanagan 2026) is the modern Skipper-coupled alternative.\n\n## Critical Decision: The Interaction-Term Design\n\n**Use `~ type + condition + type:condition` and test the interaction-term coefficient.** This is the single most consequential statistical choice in differential CLIP.\n\n- `type` = `ip` vs `sminput` (whether the library is IP or size-matched input)\n- `condition` = `treated` vs `untreated` (or KD vs control)\n- `type:condition` interaction = \"Does the IP-vs-input ratio shift across conditions?\"\n\nA naive `~ condition` design tests whether read counts differ regardless of whether they come from IP or SMInput - this confounds binding changes with expression changes. The interaction term explicitly tests for **differential binding** (the biologically meaningful signal) rather than differential expression at peak loci.\n\nWhen testing, extract the interaction-term coefficient: `results(dds, name = 'typeip.conditiontreat')`. The log2FoldChange returned is the change in IP\u002Finput ratio in `treated` vs `untreated` - this is what \"differential binding\" means.\n\n## Critical Choice: Peak-Level vs Window-Level vs Crosslink-Level\n\nThree scales differ in resolution and statistical power:\n\n**Peak-level (edgeR, limma-voom, DESeq2 on CLIPper peaks):** Test each consensus peak's IP\u002FSMInput log2 FC across conditions. Pro: peak boundaries are biologically meaningful; output interpretable. Con: peak set changes across conditions (a new peak in treatment but missing in control complicates testing); SMInput normalization must be applied consistently.\n\n**Window-level (DEWSeq, Flipper):** Tile transcriptome into fixed 50-100 nt windows; test each window. Pro: comparable across conditions (windows are pre-defined); handles binding-mode shifts within a peak; high statistical power. Con: window-resolution; multiple-testing burden (millions of windows); biological meaning of a window needs translation.\n\n**Crosslink-level (custom):** Test each single-nt CL position. Pro: nucleotide resolution; captures motif-level shifts. Con: very low coverage per position; massive multiple-testing burden; rarely used in published differential CLIP.\n\n| Goal | Scale | Tool |\n|------|-------|------|\n| ENCODE-style peak-level differential (CLIPper upstream) | Peak | DESeq2 \u002F edgeR \u002F limma-voom on CLIPper consensus peaks |\n| Maximum sensitivity windowed differential | Window | DEWSeq (with htseq-clip) or Flipper (with Skipper) |\n| Modern Skipper-coupled workflow | Window | Flipper |\n| Single-cell scCLIP differential | Cell | Specialized single-cell CLIP methods (few published tools) |\n| Compare binding-mode shifts within a peak | Window | DEWSeq |\n| Allele-specific differential | CL site | BEAPR + custom logistic |\n| RBP-KD effect on binding profile | Peak\u002Fwindow | DEWSeq (handles KD-effect on RBP itself) |\n\n## DEWSeq Workflow (Window-Level Differential)\n\nDEWSeq is the EMBL\u002FHentze-group windowed-NB framework for CLIP binding-site discovery, adapted here for differential testing via the interaction design. The pipeline is:\n\n**Goal:** Identify transcriptome windows where the IP-vs-SMInput ratio shifts across conditions, accounting for replicate variance with the negative-binomial GLM.\n\n**Approach:** Use htseq-clip to extract sliding 50 nt windows across annotated features, count reads per window per sample, build a DESeqDataSetFromSlidingWindows object with the `~ type + condition + type:condition` interaction design, extract the `typeip.conditiontreat` interaction coefficient as the differential-binding effect size, and aggregate adjacent significant windows with `bedtools merge -d 100`.\n\n```bash\n# Step 1: htseq-clip generates sliding-window count matrices\nhtseq-clip extract \\\n    -i gencode.v38.annotation.gff \\\n    -o annotation_windows.bed \\\n    --window-size 50 \\\n    --window-step 20 \\\n    --feature-type CDS,UTR\n\n# Step 2: count IP and SMInput reads per window per sample\nfor sample in ip_rep1 ip_rep2 sminput_rep1 sminput_rep2; do\n    htseq-clip count \\\n        -i ${sample}.dedup.bam \\\n        -a annotation_windows.bed \\\n        -o ${sample}.counts.txt \\\n        --mate 2\ndone\n# (For eCLIP, --mate 2 because R2 5' is the truncation site; for iCLIP single-end use --mate 1)\n\n# Step 3: DEWSeq differential testing\nhtseq-clip mergeCounts \\\n    -i ip_rep1.counts.txt ip_rep2.counts.txt sminput_rep1.counts.txt sminput_rep2.counts.txt \\\n    -o merged_counts.tsv\n```\n\n```r\nlibrary(DEWSeq)\n\ncounts \u003C- read.table('merged_counts.tsv', sep='\\t', header=TRUE, row.names=1)\ncolData \u003C- data.frame(\n    sample = c('ip_rep1','ip_rep2','sminput_rep1','sminput_rep2'),\n    type = c('ip','ip','sminput','sminput'),\n    condition = c('treated','treated','untreated','untreated')\n)\n\ndds \u003C- DESeqDataSetFromSlidingWindows(\n    countData = counts,\n    colData = colData,\n    annotObj = 'annotation_windows.bed',\n    design = ~ type + condition\n)\n\ndds \u003C- DESeq(dds)\n# In a model with `~ type + condition`, the IP-vs-input contrast is the simple condition\n# main effect; to test the differential CLIP signal between conditions use the interaction\n# term name from the design matrix (matches the skill's interaction-term guidance below):\nres \u003C- results(dds, name='typeip.conditiontreated')\n\n# Window-level FDR adjustment\nres_filtered \u003C- res[!is.na(res$padj) & res$padj \u003C 0.05 & abs(res$log2FoldChange) > 1, ]\n\n# Aggregate adjacent significant windows into differential regions\nsig_windows \u003C- as.data.frame(res_filtered)\nsig_windows$chr \u003C- gsub('_.*', '', rownames(sig_windows))\n# Custom reduce: combine adjacent windows within 100 nt\n```\n\n## Flipper Workflow (Skipper-Coupled)\n\nFlipper (Flanagan 2026) is the differential companion to Skipper, operating on the same 100 nt feature-respecting windows with a negative-binomial (DESeq2-based) differential test.\n\n```bash\n# Assume Skipper has been run on all samples; Skipper output is at skipper_out\u002F\nflipper differential \\\n    -i skipper_out\u002F \\\n    --design design.tsv \\\n    --contrast treatment vs control \\\n    -o flipper_out\u002F\n\n# design.tsv format:\n# sample_id   condition   replicate   ip_or_input\n# ip_treat_r1 treatment   1           ip\n# ip_treat_r2 treatment   2           ip\n# in_treat_r1 treatment   1           input\n# ...\n```\n\nOutput: differential window BED with log2 FC, p, padj per window.\n\n## Peak-Level Differential (CLIPper Upstream)\n\n```r\nlibrary(DESeq2)\nlibrary(GenomicRanges)\nlibrary(Rsubread)\n\n# Step 1: union of CLIPper peaks across conditions\n# (See bedtools merge upstream)\npeaks \u003C- read.table('consensus_peaks.bed', sep='\\t', col.names=c('chr','start','end','name','score','strand'))\n\n# Step 2: count reads per peak per sample with featureCounts\nsaf \u003C- data.frame(\n    GeneID = peaks$name,\n    Chr = peaks$chr,\n    Start = peaks$start + 1,  # 1-based for featureCounts\n    End = peaks$end,\n    Strand = peaks$strand\n)\ncounts_ip \u003C- featureCounts(\n    files = c('ip_treat_r1.bam','ip_treat_r2.bam','ip_ctrl_r1.bam','ip_ctrl_r2.bam'),\n    annot.ext = saf,\n    isGTFAnnotationFile = FALSE,\n    strandSpecific = 1,\n    isPairedEnd = TRUE\n)$counts\n\ncounts_in \u003C- featureCounts(\n    files = c('in_treat_r1.bam','in_treat_r2.bam','in_ctrl_r1.bam','in_ctrl_r2.bam'),\n    annot.ext = saf,\n    isGTFAnnotationFile = FALSE,\n    strandSpecific = 1,\n    isPairedEnd = TRUE\n)$counts\n\n# Step 3: DESeq2 with IP vs SMInput interaction\nall_counts \u003C- cbind(counts_ip, counts_in)\ncolData \u003C- data.frame(\n    type = rep(c('ip','input'), each=4),\n    condition = rep(c('treat','treat','ctrl','ctrl'), 2),\n    replicate = rep(c('r1','r2','r1','r2'), 2)\n)\n\ndds \u003C- DESeqDataSetFromMatrix(countData = all_counts, colData = colData,\n                               design = ~ type + condition + type:condition)\ndds \u003C- DESeq(dds)\n\n# The interaction term `typeip.conditiontreat` tests:\n# does the IP\u002Finput ratio differ in treatment vs control?\nres \u003C- results(dds, name = 'typeip.conditiontreat')\n\n# Filter\nres_sig \u003C- res[!is.na(res$padj) & res$padj \u003C 0.05 & abs(res$log2FoldChange) > 1, ]\n```\n\nThe interaction-term design (`type:condition`) is the correct statistical model for differential CLIP: it tests whether the IP-vs-input ratio differs across conditions, which is what \"differential binding\" means. Naive testing of just `condition` (ignoring SMInput) confounds binding changes with expression changes.\n\n## RBP Knockdown Experiment Design\n\nThe canonical differential CLIP design is to knock down the RBP and observe what binding sites are lost. Caveats:\n\n| Issue | Implication | Mitigation |\n|-------|-------------|------------|\n| RBP KD also depletes the RBP protein in cells | siRNA\u002FshRNA reduces RBP -> reduces IP yield -> reduces unique fragments | Normalize against SMInput WITHIN each condition; the relative IP\u002FSMInput captures binding, not protein level |\n| RBP KD changes transcript abundance | mRNA stability regulators (HuR, PUM2) when knocked down change target abundance | Both IP and SMInput see the change; ratio still works |\n| Off-target effects of siRNA | Multiple binding profiles change | Use multiple independent siRNAs; require concordance |\n| KD efficiency varies | Lower KD -> smaller binding-loss signal | Validate KD by WB on the same IP lysate; > 70% protein loss target |\n| Rescue requires re-introducing RBP | siRNA-resistant RBP cDNA for rescue | The standard differential validation experiment |\n\n## Per-Tool Failure Modes\n\n### DEWSeq -- Slow on dense libraries\n\n**Trigger:** Whole-genome window tiling at 20 nt step; dense library (50M unique fragments); 4+ samples.\n\n**Mechanism:** DEWSeq runs DESeq2 internals on millions of windows; the dispersion fit on this many features is slow.\n\n**Symptom:** Runtime > 6 h; out-of-memory; \"size of object exceeds vector limit\".\n\n**Fix:** Increase window step size to 50 nt; pre-filter windows with low counts; or restrict to expressed transcripts only. DEWSeq vignette suggests `keep \u003C- rowSums(counts(dds)) >= 30; dds \u003C- dds[keep,]` before testing.\n\n### DEWSeq -- Custom adjacency aggregation needed\n\n**Trigger:** Windows are 50 nt; biological binding sites are 50-500 nt; user expects DEWSeq to output continuous \"differential regions\" but gets individual windows.\n\n**Mechanism:** DEWSeq outputs per-window results; aggregating adjacent significant windows into regions is a separate step.\n\n**Symptom:** Output has 10,000 individual windows; user expects 1,000 biological regions.\n\n**Fix:** Use the DEWSeq utility `resultsDEWSeq()` then `bedtools merge -d 100` on the significant-window BED. Or use the `top_hits_to_bed.R` script from DEWSeq examples.\n\n### Peak-level differential -- Peak set differs between conditions\n\n**Trigger:** CLIPper called peaks separately per condition; treatment has peaks at sites missing in control (and vice versa).\n\n**Mechanism:** Peak unification requires a consensus peakset; testing on a \"treatment-only\" peak underestimates evidence in control (zero reads) and produces spurious DE.\n\n**Symptom:** \"Treatment-specific\" peaks dominate DE results; biologically implausible.\n\n**Fix:** Generate consensus peakset across all conditions (bedtools merge of all per-condition peak BEDs); count reads per consensus peak across all samples; THEN run differential. The Yeo lab convention is consensus peakset across all samples.\n\n### Interaction term forgotten\n\n**Trigger:** DESeq2 design `~ condition` instead of `~ type + condition + type:condition`.\n\n**Mechanism:** Simple `~ condition` tests whether read counts differ between treatment and control regardless of whether reads are from IP or SMInput. A condition-driven expression change in SMInput is detected as DE binding.\n\n**Symptom:** DE results dominated by transcripts with global expression changes (housekeeping shifts).\n\n**Fix:** Always use the interaction-term design. The biologically meaningful test is the interaction `type:condition` p-value.\n\n### Normalization assumptions\n\n**Trigger:** edgeR `calcNormFactors(method='TMM')` on CLIP-seq data.\n\n**Mechanism:** TMM assumes most features (genes) are not differentially expressed. CLIP-seq peak counts can be globally shifted if the RBP itself is knocked down; TMM normalization would force the shift to be invisible.\n\n**Symptom:** Knockdown experiment shows ~0 DE peaks; expected hundreds.\n\n**Fix:** Use SMInput as the control library; spike-in normalization if available; or skip TMM and use library-size normalization only. Some CLIP-specific tools (DEWSeq, Flipper) handle this internally.\n\n### Flipper requires Skipper upstream\n\n**Trigger:** Flipper called on CLIPper output.\n\n**Mechanism:** Flipper expects Skipper's window-level output format with beta-binomial estimates.\n\n**Symptom:** Flipper crashes or produces nonsense.\n\n**Fix:** Use DEWSeq with htseq-clip for CLIPper-upstream workflows; use Flipper only with Skipper.\n\n## Decision Tree by Scenario\n\n| Scenario | Tool + design | Why |\n|----------|---------------|-----|\n| KD vs control eCLIP, CLIPper upstream | DEWSeq + htseq-clip | CLIP-specific NB GLM with interaction term |\n| KD vs control eCLIP, Skipper upstream | Flipper | Skipper companion; matches windowing |\n| Treatment vs vehicle (small effect) | DEWSeq (window-level higher power) | Sliding windows capture small shifts |\n| Multiple time points | DEWSeq with time as covariate | Continuous design with time vector |\n| Allele-specific differential | BEAPR per-allele + custom logistic | See clip-seq\u002Fclip-alignment for WASP |\n| Single-cell CLIP differential | Specialized single-cell CLIP methods | Nascent; few published tools |\n| Differential motif occupancy | Window-level + DEWSeq + motif overlap | Combine differential windows with motif BED |\n| RBP overexpression vs control | Same as KD reversed | Same statistical framework |\n| Compare two RBPs | NOT differential CLIP; use SPIDR or separate CLIPs | Different RBPs need separate IPs |\n| Spike-in normalization needed | DEWSeq + spike-in size factors | For global occupancy shifts |\n| chimeric eCLIP differential miRNA targets | Custom; treat each miRNA-target chimera as feature | Specialized; see clip-seq\u002Fago-clip-mirna-targets |\n\n## Reconciliation: When Differential Tools Disagree\n\n| Pattern | Likely cause | Action |\n|---------|--------------|--------|\n| DEWSeq finds many DE windows; edgeR peak-level finds few | Window-level higher power for narrow shifts | Aggregate DEWSeq windows; cross-check |\n| edgeR many DE; DEWSeq few | edgeR not accounting for SMInput | Re-run edgeR with interaction term |\n| DE peaks dominated by expression changes | No interaction term | Use `~ type + condition + type:condition` |\n| KD experiment shows ~0 DE | TMM over-corrects global shift | Switch to library-size norm only; use SMInput |\n| siRNA replicates discordant | Off-target effects vary | Use multiple independent siRNAs; require concordance |\n| Treatment-only peaks dominate DE | No consensus peakset | Generate consensus first; then test on unified set |\n| Significant windows scattered | Window aggregation step skipped | `bedtools merge -d 100` on significant-window BED |\n| Same gene appears in many DE windows | Multiple binding sites per gene differential | Report at gene-level too; not just window-level |\n| Flipper fails with non-Skipper input | Upstream mismatch | Use DEWSeq for non-Skipper workflows |\n| DESeq2 dispersion fit fails | Too few replicates (n=2 per condition); too few features after filtering | Increase replicates; or relax filtering |\n\n**Operational rule for high-confidence differential reporting:** (a) Use SMInput-aware design (`~ type + condition + type:condition`); (b) generate consensus peakset across conditions; (c) require padj \u003C 0.05 AND |log2FC| > 1; (d) require concordance with at least one orthogonal method (e.g., DEWSeq + edgeR peak-level on same data); (e) for KD experiments, validate KD efficiency by WB and require multiple independent siRNAs.\n\n## Common Errors\n\n| Error \u002F symptom | Cause | Solution |\n|-----------------|-------|----------|\n| DESeq2 `~ condition` finds many DE | No interaction with type | Use `~ type + condition + type:condition` and test interaction |\n| DEWSeq output is gene-level not window-level | `keep \u003C- ...` filter too aggressive | Loosen pre-filter |\n| Adjacent significant windows not aggregated | Forgot bedtools merge | `bedtools merge -d 100` on sig windows |\n| edgeR TMM fits all libraries to one value | Most-features-not-DE assumption violated | Use library-size norm; or DEWSeq for CLIP-specific |\n| Flipper crashes on CLIPper input | Tool mismatch | Switch to DEWSeq |\n| Few replicates -> unstable estimates | n=2 not enough for dispersion | Increase n; or use limma-voom (more tolerant) |\n| Peaks differ across conditions | Per-condition peak calls | Unify with consensus peakset |\n| KD experiment yields 0 DE | Normalization over-corrected; OR KD efficiency too low | Validate KD WB; check normalization |\n| Global shift in IP relative to SMInput | RBP itself knocked down so IP yield lower | Normalize WITHIN each condition |\n| Lots of \"treatment-only\" peaks | Caller stringency higher in one condition | Use consensus peakset for fairness |\n\n## References\n\n- Schwarzl T et al 2024 Nucleic Acids Res 52:e1 (DEWSeq, windowed NB binding-site discovery; DESeq2-based, adaptable to differential designs)\n- Sahadevan S et al 2022 Methods Mol Biol 2404:189 (DEWSeq + htseq-clip pipeline)\n- Flanagan K, Xu S, Yeo GW 2026 bioRxiv 2026.03.13.711628 (Flipper, Skipper-companion differential; preprint)\n- Boyle EA et al 2023 Cell Genomics 3:100317 (Skipper, parent of Flipper)\n- Love MI et al 2014 Genome Biol 15:550 (DESeq2)\n- McCarthy DJ et al 2012 Nucleic Acids Res 40:4288 (edgeR)\n- Ritchie ME et al 2015 Nucleic Acids Res 43:e47 (limma)\n- Yang EW et al 2019 Nat Commun 10:1338 (BEAPR allele-specific protein-RNA binding)\n- Van Nostrand EL et al 2020 Nature 583:711 (ENCODE 150 RBP shRNA + eCLIP comparison)\n\n## Related Skills\n\n- clip-seq\u002Fclip-peak-calling - CLIPper \u002F Skipper outputs feed differential\n- clip-seq\u002Fclip-qc - Replicate QC required for valid differential\n- clip-seq\u002Fbinding-site-annotation - Annotate differential regions\n- clip-seq\u002Fclip-motif-analysis - Motif analysis on differential windows\n- clip-seq\u002Fago-clip-mirna-targets - Differential miRNA targeting from chimeric eCLIP\n- differential-expression\u002Fdeseq2-basics - Underlying NB GLM model\n- differential-expression\u002Fde-results - DE results interpretation\n- differential-expression\u002Fedger-basics - edgeR for peak counts\n- chip-seq\u002Fdifferential-binding - DNA-protein analogue\n",{"name":5,"source":24,"license":15,"tool_type":25,"description":7,"imported_at":26,"source_path":27,"primary_tool":28,"source_repository":29},"https:\u002F\u002Fgithub.com\u002FGPTomics\u002FbioSkills","r","2026-09-20","clip-seq\u002Fdifferential-clip\u002FSKILL.md","DEWSeq","GPTomics\u002FbioSkills","main","2d1750717747097b602acf30bd09a7cbd4b4c6ea","awesome_skills\u002Fbiology\u002Fbio-clip-seq-differential-clip","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous","2026-11-09 11:45:36",{"source":8,"skillName":5,"commands":37,"prompt":47,"zipAvailable":48},[38,41,44],{"tool":39,"command":40},"npx","npx skills add Mr-Nobody-Anonymous\u002FAll-skills --skill bio-clip-seq-differential-clip",{"tool":42,"command":43},"bunx","bunx skills add Mr-Nobody-Anonymous\u002FAll-skills --skill bio-clip-seq-differential-clip",{"tool":45,"command":46},"pnpm","pnpm dlx skills add Mr-Nobody-Anonymous\u002FAll-skills --skill bio-clip-seq-differential-clip","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fbiology\u002Fbio-clip-seq-differential-clip\u002FSKILL.md",true,[50,56],{"path":51,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":54,"sourceHtmlUrl":55},"examples\u002Frun_dewseq.R","OTHER","其他",3325,"https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fbiology\u002Fbio-clip-seq-differential-clip\u002Fexamples\u002Frun_dewseq.R",{"path":57,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":60,"sourceHtmlUrl":61},"usage-guide.md","REFERENCE","参考文档",4151,"https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fbiology\u002Fbio-clip-seq-differential-clip\u002Fusage-guide.md",[63,71,78,85,92,99],{"skillNo":64,"name":65,"nameZh":6,"description":66,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":67,"categories":68,"keywords":69,"publishedAt":70},"01M7FFFZPC2025JQQHPHDX154J","dei-assessment","Assess diversity, equity, and inclusion with metrics. TRIGGERS - Use when user needs help with dei-assessment related tasks.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fdei-assessment\u002FSKILL.md",[],[],"2026-11-16 10:05:34",{"skillNo":72,"name":73,"nameZh":6,"description":74,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":75,"categories":76,"keywords":77,"publishedAt":70},"01M7FFFWN8H4WA7A6Y398QKB6J","dei-action-plan","Create DEI action plans with goals and initiatives. TRIGGERS - Use when user needs help with dei-action-plan related tasks.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fdei-action-plan\u002FSKILL.md",[],[],{"skillNo":79,"name":80,"nameZh":6,"description":81,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":82,"categories":83,"keywords":84,"publishedAt":70},"01M7FFFSHTM0QZRVZ2NRJTJ0PY","culture-fit-assessment","Design culture add assessment processes. TRIGGERS - Use when user needs help with culture-fit-assessment related tasks.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fculture-fit-assessment\u002FSKILL.md",[],[],{"skillNo":86,"name":87,"nameZh":6,"description":88,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":89,"categories":90,"keywords":91,"publishedAt":70},"01M7FFFMA12RKEQA38KHJSXXPS","cowork-to-code-bridge","Use an already-installed, independently verified cowork-to-code bridge to run narrowly approved actions on the user's own macOS, Linux, or WSL2 machine through a local file queue.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fcowork-to-code-bridge\u002FSKILL.md",[],[],{"skillNo":93,"name":94,"nameZh":6,"description":95,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":96,"categories":97,"keywords":98,"publishedAt":70},"01M7FFFH1F6M8YRGW5V487GH1K","counter-offer-response","Draft counter-offer responses for retention. TRIGGERS - Use when user needs help with counter-offer-response related tasks.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fcounter-offer-response\u002FSKILL.md",[],[],{"skillNo":100,"name":101,"nameZh":6,"description":102,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":103,"categories":104,"keywords":105,"publishedAt":70},"01M7FFFDGEHV8Y7J9N40HZXW52","continuous-feedback-system","Design continuous feedback systems. TRIGGERS - Use when user needs help with continuous-feedback-system related tasks.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fcontinuous-feedback-system\u002FSKILL.md",[],[]]