ELF-NET Data S2: algaGPT-classified algal protein sequences from the mixed-source domain-analysis set

This deposit contains 2,044 FASTA files from the mixed-source ELF-NET domain-analysis set: 1,005 ocean metagenomes, 606 MMETSP transcriptomes, 236 NCBI reference genomes, and 197 other cultured reference proteomes. Each file contains proteins assigned to the algal class by algaGPT, a 70-million-parameter GPT-2/nanoGPT-based protein classifier implemented in the LA4SR framework, using top-k sampling (k = 10). algaGPT is not Pythia-based; the Pythia-70M LA4SR classifier was used only as a separate comparison model. The analysis set contains 231.7 million proteins classified as algal. Related to Figure 1, Table S1, and STAR Methods in Nelson et al., Coupling of oceanographic state to the dark proteome: a foundation for genome-informed marine productivity modeling.