Find Ardigen gene connections to pursue: exploration for MorphMap paper

[AnneCarpenter]

Ardigen and Niranj are generally providing top25 strongest gene-gene correlation and anti-correlation lists.

ARDIGEN presentation (all CRISPR based): https://docs.google.com/presentation/d/1Rnn2E-DhwLzIKPk7QwLsNmWSS0rQBOjq9LO6rVQEh80/edit#slide=id.g1ec32a8f3f1_0_4

pinged Argiden niranj to send updated lists (Anne's email link https://mail.google.com/mail/u/0/#search/tomacz/KtbxLxgVVFgnKqNbwJjPsTGBmmGxMfkFWL

Already this yielded these:
#3
#4
#5

[AnneCarpenter]

@niranjchandrasekaran emailed the info:

Here are the heatmaps. These are the top 100 gene pairs. These are also genes which are replicable.

ORF
top correlations: https://github.com/jump-cellpainting/morphmap/blob/main/12.explore-correlations-anticorrelations/figures/top_gene_pairs_clustering_transformed.png

top anticorrelations: https://github.com/jump-cellpainting/morphmap/blob/main/12.explore-correlations-anticorrelations/figures/bottom_gene_pairs_clustering_transformed.png

CRISPR
top correlations: https://github.com/zahrahanifehlou/morphmap/blob/main/12.explore-correlations-anticorrelations/figures/harmonized_CRISPR_profiles_q_thresholdtop_gene_pairs_clustering_harmonized_CRISPR_profiles.png

top anticorrelations: https://github.com/zahrahanifehlou/morphmap/blob/main/12.explore-correlations-anticorrelations/figures/harmonized_CRISPR_profiles_q_thresholdbottom_gene_pairs_clustering_transformed_harmonized_CRISPR_profiles.png

[AnneCarpenter]

@niranjchandrasekaran could you clarify what you mean by top 100? (given each chart shows fewer than 100, it must be some combo of these adds to 100)

[niranjchandrasekaran]

I generate a list of the top hundred correlated gene pairs, which many contain many occurrences of the same gene, create a list of all the unique genes in the list and then create an all-by-all heatmap. It looks like there are more unique genes in that list for the ORF data compared to the CRISPR data. I will check the code to ensure everything is working the way it should.

[tjetkaARD]

Claryfing the lists of gene pairs:

• I introduced serious misunderstanding by incorrectly embedding the figures of correlated and anti-correlated gene pairs for the the presentation for JUMP-CP meeting (attached above). I mistakenly copied or generated figures for ORFs and called them CRISPR. This propagated some misunderstanding, including the parallel discussions. I am sorry for the confusion - fully on me.
• The links to the CRISPR figures given by Niranji above indeed come from the data, but do not represent the 100 pairs. On my side, it produces much larger and not readible graph in comparison to the ORFs. I will double check with Zahra tomorrow morning and confirm the top pairs. Possibly - better to have a smaller heatmap and a separate list of top pairs.
• Once confirmed, I will follow-up with the specific hypotheses to explore & connect with the ORF results for comparison - together with full molecular background

[AnneCarpenter]

@tjetkaARD I think we can close this issue because I believe there is no longer 'evotec' vs Ardigen; it's all the same data being used and you've been working on everything we need in other issues - is that right?

No need to clean up old issues, I will just remove "evotec" and "ardigen" tags given we've converged!

[AnneCarpenter]

Tomasz confirmed in our meeting we can close this one.