Papers for

biotech development teams

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

P2p improves prediction of cell responses to gene changes

P2P: Cross-View Population Denoising for Unpaired Single-Cell Perturbation Response Prediction

Abstract: AIVC (AI Virtual Cell) is a learned simulator of cellular behavior across conditions. Predicting how a cell population responds transcriptionally to a genetic perturbation is a core task. Perturb-seq records that response by destructive sequencing, so a control cell and a perturbed cell are never observed as a pair, and cells under one condition remain heterogeneous and noisy. Regression on individual cells absorbs sampling variation into the estimated effect, whereas interpretation requires the reproducible population effect. P2P (Perturbation-to-Perturbation) takes a stochastic cell-set view as its supervision unit. Two views drawn from the same condition share a reproducible population effect and differ by view-specific variation. A permutation-invariant set encoder summarizes the control population, a structured encoder represents perturbation tokens, cellular context, dose, and combination interactions, and a gate blends empirical condition-effect memory with a neural residual. A heteroscedastic head predicts the population mean and gene-wise response variance. Under one protocol and five seeds, P2P attains the lowest expression RMSE and the highest Effect Pearson, DEG F1, and DEG average precision on each of Adamson, Norman, Replogle K562, and Replogle RPE1 relative to GenePert, LinearPert, SLIM, Scouter, and scPILOT. On Replogle K562, Effect Pearson rises from 0.643 to 0.702 and DEG F1 rises from 0.067 to 0.178 relative to Scouter, the strongest baseline on both metrics.

Mon 28 SeptEmerging TechnologiesArtificial IntelligenceMachine Learning
The gist
Predicting how groups of cells react to gene changes is important but tricky because experiments destroy cells, so you never see the same cell before and after a change. The authors developed P2P, a method that looks at groups of cells rather than individuals, capturing reliable overall effects instead of noisy single-cell data. Their approach uses special neural networks to summarize control cells and represent genetic changes, improving accuracy in predicting gene activity changes across multiple datasets. This method helps better understand how cells respond to genetic perturbations by focusing on consistent group-level patterns.
Open → 2609.34391v1

Open code enables new antibody editing method with more flexible mutations

When Edit Flows are Edit Jumps: replicating Edit Flows and EvoFlows

Abstract: Antibody lead optimization calls for a small, bounded set of edits to an existing candidate: substitutions, but also insertions and deletions. Edit-based generative models are the only ones that allocate such an edit budget without fixing the edit positions, the edit count, or the output length in advance. However, the existing approaches Edit Flows and EvoFlows did not release code or complete training specifications. Here, we show that both methods follow the same underlying process -- edits firing one at a time, at learned rates, in continuous time -- the pure-jump case of generator matching over finite sequences. With EditJumps we introduce the first open implementation of this framework, with a single generalist antibody editor trained on 1.66M Observed Antibody Space homolog pairs to propose homolog-like variants of a seed sequence, editing unseen leads zero-shot, without the per-family retraining original approaches require. Replicating this system from scratch exposes why open code is essential for generative biology: reconciling published edit distributions required reverse-engineering an undocumented rate-scaling hyperparameter that dictates realized mutation counts. Moreover, we show that published evaluation metrics are highly sensitive to reference sample size, frequently flipping method rankings. We release our full codebase, automated test suite, and configurations at: https://github.com/VisiumCH/editjumps

Wed 16 SeptMachine Learning
The gist
Improving antibodies often requires making small, precise changes to their sequences, but previous computer models had limits in how these changes could be made. The authors recreated two earlier, unpublished methods and found they work by making one change at a time at varying speeds. They introduced EditJumps, the first openly available tool that can edit antibodies in a more general way without needing retraining for each family. Their work shows that sharing code is crucial for understanding and fairly comparing methods in this field.
Open → 2609.18745v1