Interaction maps from Minerva-MLM, a genome language model for coevolutionary discovery: RNA base pairing, DNA repeats and protein contacts in one overlay.
Input can be raw DNA or FASTA (genes are called with Pyrodigal and translated), a protein, an RNA, a GenBank file, or Minerva's mixed protein + DNA format. The model reads 4,096 tokens (8,192 with the 8k checkpoint), about 10 kb (20 kb) of a typical bacterial locus.
Code and documentation: github.com/garykbrixi/minerva.
Interaction maps from Minerva-MLM, a genome language model for coevolutionary discovery: RNA base pairing, DNA repeats and protein contacts in one overlay.
Input can be raw DNA or FASTA (genes are called with Pyrodigal and translated), a protein, an RNA, a GenBank file, or Minerva's mixed protein + DNA format. The model reads 4,096 tokens (8,192 with the 8k checkpoint), about 10 kb (20 kb) of a typical bacterial locus.
Code and documentation: github.com/garykbrixi/minerva.
