From 9a1db26ce6cd8316e4092859f798a5abc0c540f1 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Artur=20Sza=C5=82ata?= Date: Thu, 19 Dec 2024 23:39:15 +0100 Subject: [PATCH] add notes to gene embedding benchmark --- _data/transformer-evaluation.yml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/_data/transformer-evaluation.yml b/_data/transformer-evaluation.yml index 55aa966..6e5a4c7 100644 --- a/_data/transformer-evaluation.yml +++ b/_data/transformer-evaluation.yml @@ -9,7 +9,7 @@ omic_modalities: Natural Language, scRNAseq, protein sequence, DNA sequence evaluated_transformers: Mistral, MPnet, CellPLM, GeneFormer, scGPT, DNABert2, ESM (plus bag-of-words, gene2vec) tasks: genomic properties, gene regulatory functions, gene localization, gene biological processes, protein properties - notes: '-' + notes: 'Evaluating model performance using gene embeddings allows comparing models utilizing different data modalities. The performance profile of a diverse set of models across ~300 gene related tasks was shown to be most influenced by the training modality.' - paper: type: preprint text: '[Ahlmann-Eltze et al. 2024](https://www.biorxiv.org/content/10.1101/2024.09.16.613342)'