We need to translate Russian news post into natural fluent English, format per rules.
First line: short headline under 90 chars, no markdown, no '#'.
Then blank line, then body split into short paragraphs (2-3 sentences each), separated by blank lines.
Wrap few genuinely important facts — key numbers, percentages, drug/company/gene names, dates — in double asterisks . At most 4-5 per post, never a whole sentence.
Wrap study/journal citations and publication references in single underscores _ (e.g., Nature Aging, July 2026). There's no explicit citation in the text; but there may be references like "Весной карта 280 биомедицинских базовых моделей раскладывала поле по типам данных и связям между ними." No citation. So maybe none.
We need to preserve all facts, numbers, names, dates exactly.
Identify key numbers: 23 August (date), 55 records, 109 models, 467 task configurations, 585 documented input routes, 280 biomedical foundation models (spring map), 40 of 55 records have several format families, 46 of 467 cases within a configuration they connect.
Also maybe "Богдан Диденко" name.
We need to wrap key numbers in double asterisks. Choose at most 4-5. Let's pick: 23 August, 55 records, 109 models, 467 task configurations, 585 documented input routes. That's 5.
Also maybe 280 biomedical foundation models but that would be 6. We need at most 4-5. So maybe we omit one. Could choose the most important: date, records, models, task configurations, input routes. That's 5.
Alternatively we could include 280 instead of something else. But we need to preserve all facts; we can still mention 280 without asterisks. That's okay.
Wrap study/journal citations in single underscores: none.
Now produce headline under 90 chars. Something like: "Bogdan Didenko releases open atlas mapping biological data routes to generative models". Count characters: Let's count: "Bogdan Didenko releases open atlas mapping biological data routes to generative models". Count: B(1) ... Let's approximate: It's likely under 90. Let's count precisely: "Bogdan Didenko releases open atlas mapping biological data routes to generative models". Characters including spaces: We'll count: Bogdan(6) + space=7, Didenko(7) =>14, space=15, releases(8)=>23, space=24, open(4)=>28, space=29, atlas(5)=>34, space=35, mapping(6)=>41, space=42, biological(10)=>52, space=53, data(4)=>57, space=58, routes(6)=>64, space=65, to(2)=>67, space=68, generative(10)=>78, space=79, models(6)=>85. So 85 chars, okay.
Now body: need short paragraphs 2-3 sentences each, separated by blank lines.
We'll translate the Russian content.
Let's craft paragraphs:
Paragraph 1: Introduce Bogdan Didenko presented open atlas of biological data routes to generative models on 23 August. The atlas shows how biological data reach generative models. In its current snapshot — 55 records, 109 models, 467 task configurations, and 585 documented input routes.
Paragraph 2: Explain generative models create new text, images, or other data; the atlas helps see what signal such a system receives. In spring a map of 280 biomedical foundation models laid out the field by data types and their connections. Didenko’s atlas looks at the next step: into what a specific task turns DNA, RNA, or text before passing it to the model.
Paragraph 3: An RNA profile — a set of gene‑activity measurements in a cell — can be fed to a model as a gene list, a character sequence encoding biological information, or a numerical vector, i.e., an ordered series of numbers. The model receives different inputs in these cases. The atlas traces the whole path: which biological object was taken, how it was transformed, and in what form the model received it.…
🔗 Read original →
Nature
Towards multimodal foundation models in molecular cell biology
Nature - The development of multimodal foundation models, pretrained on diverse omics datasets, to unravel the intricate complexities of molecular cell biology is envisioned.
August 26, 2026 3
