Comments (4)
Understood
For the sake of completeness and my future self, here's another Q I was pondering, with an obvious answer that hadn't quite sunk in until now.
In terms of profiling in general – not specific to predicting variant impact – how is the protein channel in Variant Painting different from the other channels, say ER?
By definition, the structure marked by the protein channel differs across perturbations. This is unlike with ER, where we are always observing the ER.
So when it comes to profiling with Variant Painting – the similarity between two perturbations in the protein channel (s_protein
) is conceptually different from that in the ER channels (s_er
), because
s_protein
measures the similarity across two different feature spaces (morphology of protein 1 vs morphology of protein 2), whereass_er
measures the similarity is in the same feature space (morphology of ER)
The only situation where s_protein
is conceptually the same as s_er
is when it is the same protein, and the two perturbations are variants or reference of the same gene.
So there is only one situation where it makes sense to use s_protein
: when both perturbations are of the same gene, and in this case you can use it for doing whatever you like – comparing localization, clustering, predicting impact, etc.
But when you use it to compare perturbations across genes, it's unclear what s_protein
is reporting, because the similarity is across two different spaces. At first, I thought s_protein
is a reasonable way to compare localization of two different proteins, but then (in Variant Painting) you are doing so across two different perturbation states, not the same perturbation state, and that makes things confusing. Note that what I am saying here does not negate what we've done in https://pubmed.ncbi.nlm.nih.gov/37732209/ (Fig 1 below) because although we place all genes in the same map, we are only highlighting variant mislocalizations (compared to reference of the same gene), but not comparing localizations across genes.
from 2021_09_01_varchamp.
Unasked for nitpick - your statement
But when you use it to compare perturbations across genes, it's unclear what s_protein is reporting, because the similarity is across two different spaces. At first, I thought s_protein is a reasonable way to compare localization of two different proteins, but then (in Variant Painting) you are doing so across two different perturbation states, not the same perturbation state, and that makes things confusing.
only is true if we assume/know that expression of the tagged version of the gene is itself perturbing. That is more likely to be the case in an overexpression context than a knock-in context, but even in OE, based on TA-ORF and MorphMap we definitely know not all OEs (and maybe not even MOST OEs) are.
from 2021_09_01_varchamp.
That clarifies for me - I don't think generic neg controls are useful for direct comparison to samples in this experiment other than to establish the impact of plate layout effects and other technical variation.
For both protein channel and non protein channels, we really only want to know if the variant and reference differ.
It's more OBVIOUS for the protein channel, but the facts remain the same for Non-protein channels: these channels' morphology might be impacted by the presence of the reference protein expression and therefore the reference protein is the proper comparator for the variant of interest, we cannot use generic neg controls as the reference. I think I might've been confused by this before, thinking the non-protein channels would not be affected by overexpression of the protein of interest.
from 2021_09_01_varchamp.
only is true if we assume/know that expression of the tagged version of the gene is itself perturbing. That is more likely to be the case in an overexpression context than a knock-in context, but even in OE, based on TA-ORF and MorphMap we definitely know not all OEs (and maybe not even MOST OEs) are.
This is useful to keep in mind – thanks @bethac07
from 2021_09_01_varchamp.
Related Issues (11)
- Selecting negative and positive control ORFs HOT 11
- Does low replicate correlation necessarily mean no signature? HOT 1
- poscon negcon selection experiment by Chloe HOT 9
- Processing Pipeline
- First Pass - Pilot Variant Painting data analysis HOT 30
- Cleanup repo
- Can single-cell-level classification score be used to determine variant impact? HOT 28
- Evaluation of Negative and Positive Control Selections
- Replicate MAP + null distributions
- Evaluation of well position effect
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from 2021_09_01_varchamp.