Comments (6)
Hi @notimenocall. The current ecosystem requires gene names to be the same. Thus, the query genes should be subsetted to the ones in reference otherwise set zero for the ones which are not available. For example, if you x_ref = [g1, g2, g3]
and your query has x_qeury = [g1, g3]
then the x_qeury
should change to x_qeury = [g1, 0, g3]
. More importantly, x_qeury
can not include any gene name which is not included in the reference data.
from scarches.
Thanks for your reply and clarification. This is what I have been doing now.
I am just thinking will this create bias, since the hvg are not selected based on the query data sets.
from scarches.
yes, you are right. This might induce a bias. However, the assumption is the reference data is so rich and it hast most of the signal which is needed for that organ. Incorporating new genes that were not included in reference might not make that much sense if we never had that gene and will be hard if we had this gene and the gene did not make it to final training data after feature selection. I will think about it and see if we can find an intermediate solution in later updates.
from scarches.
It makes sense. Alternatively, will it help or hurt to include more hvg genes (> 5k?) when training the reference? Thank you.
from scarches.
@notimenocall it depends on the organ or the complexity of the system you are working with, I would start with 2k and then go to 5k. For multi-tissue/species analysis higher genes are needed.
from scarches.
Good point. Thank you.
from scarches.
Related Issues (20)
- fail to use the sca.models.SCVI HOT 7
- Perform scArches in R using Seurat
- Multiple categorical_covariate_keys not allowed HOT 1
- scArches for PeakVI HOT 1
- cannot setup.py install HOT 2
- ImportError: libcudnn.so.8: cannot open shared object file: No such file or directory HOT 2
- problems about scHPL training HOT 13
- hlca_map_classify.ipynb missing early_stopping_kwarg
- Errors extracting embeddings HOT 1
- Runtime error using scPoli - mat1 and mat2 must have the same dtype, but got Double and Float HOT 7
- the difference between scPoli and scANVI HOT 1
- Update of conda environment yaml
- environment reproducibility HOT 2
- scPoli usage HOT 4
- More detailed training logs HOT 2
- scPoli Model for Unsupervised Use HOT 1
- Choice of reference and query data sets HOT 1
- Issue in annotating cell types of unlabelled query data by scPoli
- Runtime error using scPoli - Tensors must have same number of dimensions: got 1 and 2 HOT 2
- Issue with expimap model HOT 1
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 scarches.