search

microscopium / microscopium

Unsupervised clustering of high content screen samples
BSD 3-Clause "New" or "Revised" License
74 stars 22 forks source link

Potentially useful papers list #131

Open jni opened 5 years ago

jni commented 5 years ago

I'm making this issue as a way to keep track of papers that might be relevant to implement in microscopium.

New UMAP paper: https://arxiv.org/abs/1802.03426 They test UMAP on lots of different datasets, so it might have something to teach us about how to get the best results out of UMAP.

Dataset distillation: https://arxiv.org/abs/1811.10959 Summarise an image dataset by a small number of synthetic images that give accurate nearest-neighbour performance. (I'm unsure about the second part, just based on my reading of the abstract.) If it works for large images it could be very good indeed for us.

jni commented 4 years ago

Note also that UMAP has a parameter for supervised/semi-supervised clustering: https://umap-learn.readthedocs.io/en/latest/supervised.html

This would be extremely useful for taking replicates into account in the embedding.

Another relevant paper: "Domain-invariant features for mechanism of action prediction in a multi-cell-line drug screen" https://academic.oup.com/bioinformatics/article/36/5/1607/5586889

jni commented 3 years ago

Cell Morphology-Guided De Novo Hit Design by Conditioning Generative Adversarial Networks on Phenotypic Image Features

https://chemrxiv.org/articles/Cell_Morphology-Guided_De_Novo_Hit_Design_by_Conditioning_Generative_Adversarial_Networks_on_Phenotypic_Image_Features/11594067

jni commented 3 years ago

The "lead generation" and "identifying the MoA" sections in this new review from the Carpenter group recapitulate the motivations behind microscopium.

Image-based profiling for drug discovery: due for a machine-learning upgrade?