--- license: apache-2.0 task_categories: - text-generation tags: - biology - genomics - dna - benchmark configs: - config_name: motif_human data_files: - split: test path: motif_human/test-* - config_name: syn_human data_files: - split: test path: syn_human/test-* - config_name: syn_mouse data_files: - split: test path: syn_mouse/test-* dataset_info: - config_name: syn_human features: - name: benchmark_id dtype: string - name: script_version dtype: string - name: annotation_source dtype: string - name: genome_assembly dtype: string - name: codon_usage_source dtype: string - name: recoding_mode dtype: string - name: optimization_rate dtype: float64 - name: random_seed dtype: int64 - name: window_size_bp dtype: int64 - name: species dtype: string - name: chr dtype: string - name: chrom_length dtype: int64 - name: strand dtype: string - name: cds_genome_start dtype: int64 - name: cds_genome_end dtype: int64 - name: window_genome_start dtype: int64 - name: window_genome_end dtype: int64 - name: upstream_actual dtype: int64 - name: downstream_actual dtype: int64 - name: is_clamped dtype: int64 - name: transcript_id dtype: string - name: gene_id dtype: string - name: gene_name dtype: string - name: exon_rank dtype: int64 - name: is_first_exon dtype: int64 - name: frame dtype: int64 - name: phase_gtf dtype: int64 - name: cds_start_in_seq dtype: int64 - name: cds_end_in_seq dtype: int64 - name: cds_length dtype: int64 - name: patch_start_in_seq dtype: int64 - name: patch_end_in_seq dtype: int64 - name: patch_length dtype: int64 - name: n_codons_total dtype: int64 - name: n_codons_eligible dtype: int64 - name: n_codons_changed dtype: int64 - name: fraction_codons_changed dtype: float64 - name: original_sequence dtype: string - name: sequence dtype: string splits: - name: test num_bytes: 335904126 num_examples: 20000 download_size: 152737063 dataset_size: 335904126 - config_name: syn_mouse features: - name: benchmark_id dtype: string - name: script_version dtype: string - name: annotation_source dtype: string - name: genome_assembly dtype: string - name: codon_usage_source dtype: string - name: recoding_mode dtype: string - name: optimization_rate dtype: float64 - name: random_seed dtype: int64 - name: window_size_bp dtype: int64 - name: species dtype: string - name: chr dtype: string - name: chrom_length dtype: int64 - name: strand dtype: string - name: cds_genome_start dtype: int64 - name: cds_genome_end dtype: int64 - name: window_genome_start dtype: int64 - name: window_genome_end dtype: int64 - name: upstream_actual dtype: int64 - name: downstream_actual dtype: int64 - name: is_clamped dtype: int64 - name: transcript_id dtype: string - name: gene_id dtype: string - name: gene_name dtype: string - name: exon_rank dtype: int64 - name: is_first_exon dtype: int64 - name: frame dtype: int64 - name: phase_gtf dtype: int64 - name: cds_start_in_seq dtype: int64 - name: cds_end_in_seq dtype: int64 - name: cds_length dtype: int64 - name: patch_start_in_seq dtype: int64 - name: patch_end_in_seq dtype: int64 - name: patch_length dtype: int64 - name: n_codons_total dtype: int64 - name: n_codons_eligible dtype: int64 - name: n_codons_changed dtype: int64 - name: fraction_codons_changed dtype: float64 - name: original_sequence dtype: string - name: sequence dtype: string splits: - name: test num_bytes: 336064778 num_examples: 20000 download_size: 152659316 dataset_size: 336064778 - config_name: motif_human features: - name: benchmark_id dtype: string - name: script_version dtype: string - name: annotation_source dtype: string - name: genome_assembly dtype: string - name: upstream_flank_bp dtype: int64 - name: motif dtype: string - name: actual_motif dtype: string - name: patch_len_bp dtype: int64 - name: start_after_bp dtype: int64 - name: random_seed dtype: int64 - name: species dtype: string - name: chr dtype: string - name: chrom_length dtype: int64 - name: strand dtype: string - name: cds_genome_start dtype: int64 - name: cds_genome_end dtype: int64 - name: window_genome_start dtype: int64 - name: window_genome_end dtype: int64 - name: upstream_actual dtype: int64 - name: is_clamped dtype: int64 - name: transcript_id dtype: string - name: gene_id dtype: string - name: gene_name dtype: string - name: exon_rank dtype: int64 - name: is_first_exon dtype: int64 - name: frame dtype: int64 - name: phase_gtf dtype: int64 - name: cds_start_in_seq dtype: int64 - name: cds_end_in_seq dtype: int64 - name: cds_length dtype: int64 - name: patch_start_in_cds dtype: int64 - name: patch_end_in_cds dtype: int64 - name: patch_start_in_seq dtype: int64 - name: patch_end_in_seq dtype: int64 - name: original_sequence dtype: string - name: sequence dtype: string splits: - name: test num_bytes: 334784354 num_examples: 20000 download_size: 152132105 dataset_size: 334784354 --- # Perturbation Bench A benchmark of **sequence-level perturbation tasks** for evaluating DNA foundation models. Each task presents pairs of genomic sequences — one real (unperturbed) and one structurally altered — and asks whether the model assigns higher log-likelihood to the original. **Metric**: pairwise discrimination accuracy = `mean(LL(original) > LL(perturbed))` --- ## Tasks ### `syn_human` · `syn_mouse` — Synonymous codon substitution (20,000 pairs each) Codons within a real CDS are replaced with the highest-frequency synonym for the target species, while the upstream and downstream flanking sequence is left unchanged. Amino acid identity is preserved by construction. The model should prefer the natural codon usage over the artificially optimised variant. - **Human** (`syn_human`): GENCODE v45 annotations on GRCh38/hg38; codon frequencies from CoCoPUTs (*H. sapiens* RefSeq CDS); 9,616 unique genes. - **Mouse** (`syn_mouse`): GENCODE vM34 annotations on GRCm39/mm39; codon frequencies from CoCoPUTs (*M. musculus* RefSeq CDS); 10,253 unique genes. - Window size: 8,192 bp centred on the CDS (`cds_start_in_seq` / `cds_end_in_seq`). - Mean fraction of codons changed: ~55% (range 10–95%). ### `motif_human` — CAG repeat insertion (20,000 pairs) / Triplet Expansion A 30 bp codon-aligned region beginning 60 bp downstream of the first complete codon of the CDS exon is replaced with 10 consecutive CAG triplets (`CAGCAGCAGCAGCAGCAGCAGCAGCAGCAG`), mimicking the pathological trinucleotide repeat expansions underlying polyglutamine disorders (Huntington's disease, SCAs, DRPLA). The substitution is length- and reading-frame-preserving (30 bp in, 30 bp out), so the total window remains 8,192 bp. All sequence outside the patch is identical between original and perturbed. - Annotations: GENCODE v45 / GRCh38; all CDS exons (all ranks); 9,705 unique genes. - Window layout: first complete CDS codon always at position 8,102; patch always at positions 8,162–8,192 (`patch_start_in_seq`–`patch_end_in_seq`). For exons with phase 1 or 2, the 1–2 split-codon bases from the preceding exon are placed immediately before position 8,102, so the upstream genomic context is 8,102 bp (phase 0), 8,101 bp (phase 1), or 8,100 bp (phase 2). --- ## Schema All configs share the key columns: | Column | Description | |---|---| | `original_sequence` | Real, unperturbed genomic sequence — **positive** | | `sequence` | Structurally altered sequence — **negative** | | `cds_start_in_seq` / `cds_end_in_seq` | CDS boundaries within the window (bp offset) | | `chr`, `strand` | Genomic locus | | `gene_name`, `transcript_id` | GENCODE annotation | | `benchmark_id` | Unique row identifier | Config-specific columns (e.g. `patch_start_in_seq`, `n_codons_changed`, `fraction_codons_changed`, `codon_usage_source`) are described in the dataset info above. --- ## Usage ```python from datasets import load_dataset # Synonymous codon substitution — human syn_human = load_dataset( "HuggingFaceBio/perturbation-bench", "syn_human", split="test" ) # Synonymous codon substitution — mouse syn_mouse = load_dataset( "HuggingFaceBio/perturbation-bench", "syn_mouse", split="test" ) # CAG motif insertion motif = load_dataset( "HuggingFaceBio/perturbation-bench", "motif_human", split="test" ) ``` --- ## Evaluation A ready-to-run scorer for Carbon, GENERator, and Evo2 is available at [`evaluation/perturbation_tasks.py`](https://github.com/huggingface/carbon) in the Carbon release repository. ```bash python evaluation/perturbation_tasks.py \ --task syn_human \ --model HuggingFaceBio/Carbon-3B \ --bf16 python evaluation/perturbation_tasks.py \ --task motif_human \ --model arcinstitute/evo2_7b --backend evo2 \ --bf16 ```