Fig. 3: Distributions of LookingGlass embeddings across environmental packages. | Nature Communications

Fig. 3: Distributions of LookingGlass embeddings across environmental packages.

From: Deep learning of a bacterial and archaeal universal language of life enables transfer learning and illuminates microbial dark matter

Fig. 3

a Pairwise cosine similarity (in red) among the average embeddings of 20,000 randomly selected sequences from each environmental package. b t-SNE visualization of the embedding space for 20,000 randomly selected sequences from each of ten distinct environmental contexts in the “mi-faser functional” validation set. Sequences from the same environmental context generally cluster together. Colors indicate environmental package. Embeddings are significantly differentiated by environmental package (MANOVA P < 10−16). Source data are provided as a Source Data file.

Back to article page