CFdb: feature tables

This upload contains features that were used to derive interactomes based on meta-analysis of co-fractionation mass spectrometry (CF-MS) data. Features reflect the similarity of any two proteins' elution profiles in a given CF-MS experiment. They were provided as input to machine-learning models trained in cross-validation on known protein complexes. Feature tables are provided here (1) to enable re-analysis of this data and (2) to enhance the analysis of future individual datasets by transfer learning.

Each .tar file contains feature tables for a given organism. In these tables, each row represents a unique protein pair and each column represents the value of that feature in a given CF-MS experiment. The top-five best-performing features identified in our previous meta-analysis (doi: 10.1038/s41592-021-01194-4) are pre-calculated for each organism, including the distance correlation, weighted cross-correlation, cosine similarity, mutual information, and Pearson correlation. Each feature was calculated using the optimal chromatogram preprocessing strategies identified in our previous meta-analysis.

Feature tables are provided for the following organisms:

Anabaena sp. PCC 7120

Arabidopsis thaliana

Brassica oleracea

Caenorhabditis elegans

Chaetomium thermophilum

Chlamydomonas reinhardtii

Cyanothece ATCC 51142

Dictyostelium discoideum

Drosophila melanogaster

Escherichia coli

Glycine max

Gossypium hirsutum

Homo sapiens

Kuenenia stuttgartiensis

Mus musculus

Nematostella vectensis

Oryza sativa

Plasmodium berghei

Plasmodium falciparum

Plasmodium knowlesi

Podospora anserina

Rattus norvegicus

Saccharomyces cerevisiae

Salmonella typhimurium SL1344

Selaginella moellendorffii

Solanum lycopersicum

Strongylocentrotus purpuratus

Synechocystis sp. PCC 6803

Triticum aestivum

Trypanosoma brucei

Xenopus laevis

Zea mays