{
  "id": 1805200,
  "title": "INDELVAR: structure-informed prediction of in-frame indel pathogenicity with calibrated PP3/BP4 thresholds",
  "url": "https://urgent.news/2026/08/18/indelvar-structure-informed-prediction-of-in-frame-indel",
  "topic": "science",
  "section": "Science",
  "published": "2026-08-18T00:00:00.000Z",
  "source": {
    "name": "bioRxiv",
    "slug": "biorxiv",
    "url": "https://www.biorxiv.org/content/10.64898/2026.08.13.737497v1?rss=1"
  },
  "original_language": "en",
  "account": "In-frame insertions and deletions are challenging to interpret due to the dependence of their effects on both the sequence change and the protein context. Researchers have created INDELVAR, a random forest model designed to predict the pathogenicity of these in-frame variations. This model incorporates 37 features, including AlphaFold-derived structural information, evolutionary conservation, local sequence alterations, gene constraint, and protein annotations.\n\nPathogenic variants tend to impact protein regions characterized by high AlphaFold confidence, limited solvent exposure, dense local packing, and robust evolutionary conservation. INDELVAR demonstrated exceptional performance in cross-validation, with an area under the receiver operating characteristic curve (AUROC) of 0.980. In an independent test set, the model achieved an AUROC of 0.977. INDELVAR outperformed other evaluated methods for both deletions and insertions, although the differences from a protein language model-based method were not statistically significant.\n\nBy separately calibrating the model for deletions and insertions, INDELVAR achieved strong evidence for both pathogenic and benign variants within each type. This range of evidence has not been previously reported for an in-frame indel predictor. During independent testing, all evidence intervals met their respective likelihood ratio requirements. A precomputed resource is available, providing scores for 372,090 observed in-frame indels mapped to Genome Reference Consortium Human Build 38.",
  "summary": "In-frame insertions and deletions are difficult to interpret because their effects depend on both the sequence change and its protein context. We developed INDELVAR, a random forest model for in-frame insertions and deletions of 1-10 amino acids that integrates 37 features describing AlphaFold-derived wild-type structural context, evolutionary conservation, local sequence change, gene constraint,…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}