PHILHARMONIC Results for P. damicornis, C. goreaui, and D. melanogaster

Overview

PHILHARMONIC is a novel computational approach that couples deep learning de novo network inference with robust unsupervised spectral clustering algorithms to uncover functional relationships and high-level organization in non-model organisms. Our clustering approach allows us to de-noise the predicted network, producing highly informative functional modules.

We apply PHILHARMONIC to predict clusters with significant functional coherence in the reef-building coral P. damicornis and its algal symbiont C. goreaui, and the well-annotated fruit fly D. melanogaster.

Description of Files

• `_sequences.fasta`: Contains amino acid sequences for each protein in the network, used as the starting point for PHILHARMONIC.
• `_GO_map.csv`: Contains the full list of Pfam domains and GO terms assigned to each protein by hmmscan and GoDomainMiner.
• `_network_positive.tsv`: 3-column table, where the first two columns are a pair of proteins, and the third is the D-SCRIPT score. Filtered only to pairs with score >= 0.5.
• `_clusters.json`: Structured data for each cluster, containing a unique hash key, the list of proteins, the predicted edges between them, and the count of GO terms associated with the cluster.
• `_clusters.llm.json`: The same as the previous, but with additional LLM-derived annotations.
• `_cluster_graph.tsv`: 3-column table, where the first two columns are a pair of clusters (indicated by the unique hash key provided in the json file), and the third is the number of crossing edges between proteins in each cluster.
• `_cluster_graph_functions.tsv`: 3-column table, containing clusters by hash key, the number of proteins in the cluster, and the GO-slim term assigned to that cluster.
• `_protein_extended_data.csv`: A protein-centric table, containing diverse information such as sequence, domain and GO annotations, and cluster assignment.
• `_human_readable.txt`: An easily-scannable file with plain text descriptions of each cluster, including summaries of their size, connectivity, and top functional terms.