arXiv subjects

Charalampos Koilakos

Publications and source records attributed to Charalampos Koilakos.

3 recordsLinked to original sources

arXiv2026

Poisoning the corpora of protein design systems during pre-training and retrieval

Protein foundation models increasingly support the generation of novel biological sequences, yet their reliance on large public datasets creates opportunities for adversarial manipulation of the material they learn from and retrieve. We investigate two data-only poisoning attacks against protein design systems, targeting the pre-training corpus of autoregressive protein language models and the retrieval corpus of a text-guided protein generator. For the first, we pre-trained ProtGPT2 and ProGen2 variants of 300 million parameters, introducing poisoned protein sequences containing one of two triggers followed by one of three payloads. Across all poisoned runs, both architectures acquired highly reliable trigger-dependent generation under uniform and ramped poisoning schedules. Under ramped exposure, the earliest majority activation occurred after only 200 contaminated examples, while for the uniform schedule earliest activation occurred around 5000 poisoned samples. For the second attack, we inserted records into a retrieval corpus system without modifying its weights or decoding procedure. A single query-aware record captured a top-16 retrieval position for 58\% of templated requests and 97\% of free-text requests. Conditional on successful poisoned retrieval, 62\% of requests produced altered sequences, while 19-21% acquired the inserted domain by profile-HMM homology. A universal query-unaware attack using 1,500 records also achieved substantial retrieval capture and domain transfer. Together, these findings demonstrate that adversarial protein sequence contributions can establish persistent, sequence-specific behaviour during pre-training or redirect protein generation at inference, while leaving ordinary model behaviour largely preserved. Our work establishes the integrity and biological provenance of protein sequence data as critical safeguards for generative protein design systems.

cs.LG↗

arXiv2026

Poisoning the Genome: Targeted Backdoor Attacks on DNA Foundation Models

Foundation models trained on DNA sequences have achieved strong performance across biological tasks including variant effect prediction and genome design. These models rely on massive public genomic datasets comprising trillions of nucleotide tokens. Unlike natural language, DNA sequences lack semantic transparency, making corrupted or adversarially crafted entries difficult to detect during data curation. We present the first systematic study of training data poisoning in genomic language models, targeting both pre-training and fine-tuning stages. At pre-training, using Evo 2 and GENERator architectures, we show that less than 1% adversarially crafted sequences in the training corpus can selectively degrade generative performance on targeted genomic contexts while leaving unrelated sequences unaffected. We evaluate three scenarios: corruption of TATA-box promoter motifs, disruption of CTCF binding sites, and insertion of synthetic sequences absent from all training genomes. At fine-tuning, we demonstrate two additional attacks. First, poisoning a subset of CTCF sites in a ClinVar-derived corpus installs a conditional backdoor in a LoRA-adapted model that activates almost exclusively when the trigger sequence is present. Second, using frozen Evo 2 7B embeddings, targeted label corruption of downstream training data selectively compromises a clinically relevant variant classification task, demonstrated on BRCA1 variant effect prediction. These results show genomic foundation models are susceptible to targeted data poisoning with minimal footprint. We urge the field to adopt data provenance tracking, integrity verification, and adversarial robustness evaluation as standard components of the genomic model development pipeline.

q-bio.GN↗

arXiv2025

Investigating DNA words and their distributions across the tree of life

The frequency distributions of DNA k-mers are shaped by fundamental biological processes and offer a window into genome structure and evolution. Inspired by analogies to natural language, prior studies have attempted to model genomic k-mer usage using Zipf's law, a rank-frequency law originally formulated for words in human language. However, the extent to which this law accurately captures the distribution of k-mers across diverse species remains unclear. Here, we systematically analyze k-mer frequency spectra across more than 225,000 genome assemblies spanning all three domains of life and viruses. We demonstrate that Zipf's law consistently underperforms in modeling k-mer distributions. In contrast, we propose the truncated power law and Zipf-Mandelbrot distributions, which provide substantially improved fits across taxonomic groups. We show that genome size and GC content influence model performance, with larger and GC-content imbalanced genomes yielding better adherence. Additionally, we perform an extensive analysis on vocabulary expansion and exhaustion across the same organisms using Heaps' law. We apply our modeling framework to evaluate simulated genomes generated by k-let preserving shuffling and deep generative language models. Our results reveal substantial differences between organismal genomes and their synthetic or shuffled counterparts, offering a novel approach to benchmark the biological plausibility of artificial genomes. Collectively, this work establishes new standards for modeling genomic k-mer distributions and provides insights relevant to synthetic biology, and evolutionary sequence analysis.

q-bio.GN↗