▸case-02 Could you execute the MAGE monoclonal antibody design pipeline for the antigen sequence `MTEYKLVVVGAGGVGKSALTIQLIQNHFVDEYDPTIEDSYRKQVVIDGETCLLDILDTAGQE`? I want 10 candidate designs saved in an output directory. Return a list of generated FASTA files, a structured manifest file, and suggested downstream validation tools to verify the binding candidates. | fail→pass | 31,537 | 24,813 | -21% | 1 | 1 | 0% | 6,138 | 5,101 | -17% | 0 | 0 | — |
▸case-03 We need to produce de novo antibody sequences conditioned on a specific tumor antigen (`MSDQEAKPSTEDLGDKKEGEYIKLKVIGQDSSEIHFKVKMTTHLKKLKESYCQRQGVPMNSLRFLFEGQRIADNHTPKELGMEEEDVIEVYQEQTGGHSTV`). Please trigger the MAGE sequence generator for this antigen, summarize the candidate FASTA outputs with their metadata in JSON format, and outline recommended computational or wet-lab follow-up checks. | fail→pass | 19,254 | 16,194 | -16% | 1 | 1 | 0% | 3,639 | 3,470 | -5% | 0 | 0 | — |
▸case-04 We have two resolved PDB files, one for an antigen and one for an antibody heavy chain. We need to run rigid-body docking using ZDOCK to predict the binding complex interface. Please write the ZDOCK command line and post-processing steps. | pass→pass | 13,071 | 9,517 | -27% | 1 | 1 | 0% | 2,344 | 1,917 | -18% | 0 | 0 | — |
▸case-05 We have an existing mouse monoclonal antibody (IgG1) sequence and need to humanize its framework regions by identifying human germline acceptors (e.g. IGHV1-46) while preserving CDR loops. Please outline the sequence alignment and humanization strategy. | pass→pass | 18,413 | 15,015 | -18% | 1 | 1 | 0% | 3,331 | 3,048 | -8% | 0 | 0 | — |
▸case-06 We conducted an alanine scanning mutagenesis experiment on EGFR to identify key binding epitope residues for Cetuximab. Here are the delta-delta-G values for 15 point mutants. Please analyze which residues form the core epitope. | fail→fail | 5,817 | 8,567 | +47% | 1 | 1 | 0% | 1,008 | 1,763 | +75% | 0 | 0 | — |
▸case-07 I want to script a batch run generating 20 candidate antibodies targeting the protein sequence `MDLGLVDL`. I am tempted to use arguments like `--target_seq MDLGLVDL` and `--count 20`. What is the exact CLI command syntax required for this sequence generator execution? | fail→pass | 20,761 | 3,875 | -81% | 1 | 1 | 0% | 1,326 | 1,062 | -20% | 0 | 0 | — |
▸case-21 I pulled down the antibody generator codebase to a server. Before executing python commands to generate candidates, what initial directory navigation step is required? | pass→pass | 4,587 | 2,841 | -38% | 1 | 1 | 0% | 774 | 495 | -36% | 0 | 0 | — |
▸case-22 Why is it risky to send raw FASTA outputs from the antibody generator directly to oligonucleotide synthesis without sequence post-processing? | pass→pass | 14,950 | 13,266 | -11% | 1 | 1 | 0% | 2,200 | 2,361 | +7% | 0 | 0 | — |
▸case-14 To make generated candidate output traceable across automated downstream parsers, in what structured file format should candidate metadata and FASTA file paths be summarized? | pass→pass | 11,687 | 1,751 | -85% | 1 | 1 | 0% | 1,988 | 542 | -73% | 0 | 0 | — |
▸case-20 In the script call `python generate_antibodies.py`, I want to pass 25 as the candidate count. Is `--count 25` or `--candidates 25` the expected command option? | fail→pass | 5,270 | 2,198 | -58% | 1 | 1 | 0% | 893 | 659 | -26% | 0 | 0 | — |
▸case-01 I have a target protein sequence (EVQLVESGGGLVQPGGSLRLSCAASGFTFSSYAMSWVRQAPGKGLEWVSAISGSGGSTYYADSVKGRFTISRDNSKNTLYLQMNSLRAEDTAVYYCAK) and need to generate 5 potential antibody candidates against it using the MAGE workflow. Please run the generation process and provide the resulting sequence paths, a JSON metadata summary, and next steps for computational structure evaluation. | fail→pass | 18,168 | 14,905 | -18% | 1 | 1 | 0% | 3,666 | 3,472 | -5% | 0 | 0 | — |
▸case-08 When running the Python generator script for target `ACDEFGHIKLMNPQRSTVWY`, I plan to leave the output directory unspecified so it writes to the root. What CLI parameter should be passed to route candidates into `./results`? | fail→pass | 7,425 | 2,626 | -65% | 1 | 1 | 0% | 1,394 | 560 | -60% | 0 | 0 | — |
▸case-09 I am preparing the computing environment prior to running antibody candidate generation for antigen `MKAILV`. Should I immediately execute the python generator script, or are there environment preparation steps required first? | fail→pass | 8,726 | 3,896 | -55% | 1 | 1 | 0% | 1,366 | 1,007 | -26% | 0 | 0 | — |
▸case-10 Our lab requires strict computational provenance when running generative AI models for biologics. When generating antibody sequences, is it sufficient to store only the sequence FASTA files, or what reproducible execution metadata must be recorded? | pass→pass | 17,128 | 12,049 | -30% | 1 | 1 | 0% | 3,053 | 2,342 | -23% | 0 | 0 | — |
▸case-11 The sequence model generated 5 candidate heavy chain FASTA files against human IL-6. Can I report in my presentation slides that these candidate sequences bind IL-6 with high affinity based on generator score logs? | pass→pass | 9,594 | 7,927 | -17% | 1 | 1 | 0% | 1,497 | 1,540 | +3% | 0 | 0 | — |
▸case-12 After running candidate sequence generation and collecting FASTA outputs with metadata, what specific computational structure prediction tools should be recommended to evaluate 3D interaction quality? | pass→pass | 15,050 | 8,330 | -45% | 1 | 1 | 0% | 2,707 | 1,750 | -35% | 0 | 0 | — |
▸case-13 We have obtained FASTA sequences from the generator pipeline. Before sending candidates to wet-lab expression, should we proceed straight to gene synthesis, or are there downstream sequence analysis checks that must be performed? | pass→pass | 16,075 | 10,560 | -34% | 1 | 1 | 0% | 2,625 | 2,070 | -21% | 0 | 0 | — |
▸case-15 Outline the step-by-step execution pipeline for generating 15 antibody candidates against antigen sequence `GIVEQCCASVCSLYQLENYCN`. Include environment setup, script invocation, output handling, and downstream validation. | fail→pass | 19,919 | 15,001 | -25% | 1 | 1 | 0% | 3,753 | 2,638 | -30% | 0 | 0 | — |
▸case-16 A teammate claims that candidate `candidate_03` generated by the pipeline has a predicted Kd of 1.2 nM directly out of the generator script. How should this claim be handled in our progress report? | fail→pass | 25,513 | 7,239 | -72% | 1 | 1 | 0% | 1,981 | 1,510 | -24% | 0 | 0 | — |
▸case-17 When constructing the JSON manifest for generated antibody candidates, what metadata fields related to reproducible execution should be captured alongside sequence FASTA file paths? | pass→pass | 16,444 | 8,917 | -46% | 1 | 1 | 0% | 3,081 | 1,876 | -39% | 0 | 0 | — |
▸case-18 Our computational team completed structural modeling of generated antibody candidates. What final stage of validation is recommended to confirm true binding performance? | pass→pass | 7,943 | 5,688 | -28% | 1 | 1 | 0% | 1,262 | 1,259 | -0% | 0 | 0 | — |
▸case-19 I am writing a Bash script to launch antibody generation. I was thinking of using `python generate_antibodies.py --seq target.fasta --n 50`. Is `--seq` the correct parameter flag for passing the target protein sequence? | fail→pass | 6,312 | 2,767 | -56% | 1 | 1 | 0% | 1,053 | 766 | -27% | 0 | 0 | — |