Work

Benchmark-Optimized Pipeline Identifies Experimentally Validated Antifreeze Peptides in Antarctic Krill

Year: 2026
Type: article
Abstract: Antifreeze proteins (AFPs) offer promising prospects as biocryoprotectants, yet their discovery remains a great challenge due to the lack of efficient large-scale screening tools. Here, we present a benchmark-optimized, multimodel computational pipeline that integrates property evaluation and structural assessment for high-throughput AFP identification. Through systematic benchmarking on independent testing data set, two optimal strategies were identified: the two-model-based Strategy 8 achieved the highest accuracy of 0.82, while the stricter four-model-based Strategy 18 attained the highest precision of 0.97, making it optimal for initial large-scale screening. Furthermore, this pipeline was applied to screen 153,359 Antarctic krill sequences and discovered 13 novel antifreeze peptides. Dynamic ice shaping assays revealed that four out of 13 peptides significantly inhibited ice crystal growth. Notably, Peptide 10 (SPHSWANTYQQCK) could inhibit ice growth and recrystallization. This study establishes a robust framework for large-scale AFP identification and provides a set of experimentally validated Antarctic krill peptides with strong potential as novel biobased cryoprotectants.
Language: English
Open Access status: green
OpenAlex ID: W4415313995
FWCI:
Cites: 76
Cited by:
Related to: 10