CellBender version 0.3.1¶
2023-11-08 16:47:44
Brain_10X.h5¶
Loaded dataset¶
AnnData object with n_obs × n_vars = 50000 × 32285
obs: 'background_fraction', 'cell_probability', 'cell_size', 'droplet_efficiency', 'n_raw', 'n_cellbender'
var: 'ambient_expression', 'feature_type', 'genome', 'gene_id', 'cellbender_analyzed', 'n_raw', 'n_cellbender'
uns: 'cell_size_lognormal_std', 'empty_droplet_size_lognormal_loc', 'empty_droplet_size_lognormal_scale', 'swapping_fraction_dist_params', 'estimator', 'features_analyzed_inds', 'fraction_data_used_for_testing', 'learning_curve_learning_rate_epoch', 'learning_curve_learning_rate_value', 'learning_curve_test_elbo', 'learning_curve_test_epoch', 'learning_curve_train_elbo', 'learning_curve_train_epoch', 'target_false_positive_rate'
obsm: 'cellbender_embedding'
layers: 'raw', 'cellbender'
Examine how many counts were removed in total¶
removed 3613126 counts from non-empty droplets removed 7.88% of the counts in non-empty droplets
Rough estimate of expectations based on nothing but the plot above: roughly 4638392 noise counts should be in non-empty droplets that is approximately 10.12% of the counts in non-empty droplets with a false positive rate [FPR] of 1.0%, we would expect to remove about 11.12% of the counts in non-empty droplets
The algorithm removed a bit less than naive expectations would indicate, but this is likely okay. If removal seems insufficient, the FPR can be increased.
Assessing convergence of the algorithm¶
The learning curve tells us about the progress of the algorithm in inferring all the latent variables in our model. We want to see the ELBO increasing as training epochs increase. Generally it is desirable for the ELBO to converge at some high plateau, and be fairly stable.
What to watch out for:
1. large downward spikes in the ELBO (of value more than a few hundred) 2. the test ELBO can be smaller than the train ELBO, but generally we want to see both curves increasing and reaching a stable plateau. We do not want the test ELBO to dip way back down at the end. 3. lack of convergence, where it looks like the ELBO would change quite a bit if training went on for more epochs.
Automated assessment --------
- WARNING: The training ELBO deviates quite a bit from the max value during the second half of training.
- We typically expect to see the training ELBO increase almost monotonically. This curve seems to have a concerted period of motion in the wrong direction near epoch 40. If this is early in training, this is probably okay.
Summary:
This is unusual behavior, and a reduced --learning-rate is indicated. Re-run with half the current learning rate and compare the results.
Examine count removal per gene¶
Pearson correlation coefficient for the above is 0.9582
This meets expectations.
Table of top genes removed¶
Ranked by fraction removed, and excluding genes with fewer than 3535 total raw counts (90th percentile)
| ambient_expression | feature_type | genome | gene_id | cellbender_analyzed | n_raw | n_cellbender | n_removed | fraction_removed | fraction_remaining | n_raw_cells | n_cellbender_cells | n_removed_cells | fraction_removed_cells | fraction_remaining_cells | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| gene_name | |||||||||||||||
| Tmem131 | 0.000048 | Gene Expression | mm10 | ENSMUSG00000026116 | True | 4294 | -115611 | 119905 | 27.923847 | -26.923847 | 3658 | -115611 | 119269 | 32.604975 | -31.604975 |
| Smap1 | 0.000071 | Gene Expression | mm10 | ENSMUSG00000026155 | True | 4810 | -27997 | 32807 | 6.820582 | -5.820582 | 3859 | -27997 | 31856 | 8.254988 | -7.254988 |
| Arhgef4 | 0.000080 | Gene Expression | mm10 | ENSMUSG00000037509 | True | 3923 | -18782 | 22705 | 5.787662 | -4.787662 | 2812 | -18782 | 21594 | 7.679232 | -6.679232 |
| Prex2 | 0.000036 | Gene Expression | mm10 | ENSMUSG00000048960 | True | 3868 | -16362 | 20230 | 5.230093 | -4.230093 | 3344 | -16362 | 19706 | 5.892943 | -4.892943 |
| Atp6v1h | 0.000054 | Gene Expression | mm10 | ENSMUSG00000033793 | True | 5252 | -21867 | 27119 | 5.163557 | -4.163557 | 4495 | -21867 | 26362 | 5.864739 | -4.864739 |
| Ogfrl1 | 0.000122 | Gene Expression | mm10 | ENSMUSG00000026158 | True | 4604 | -18454 | 23058 | 5.008254 | -4.008254 | 2906 | -18454 | 21360 | 7.350310 | -6.350310 |
| Gdap1 | 0.000064 | Gene Expression | mm10 | ENSMUSG00000025777 | True | 3967 | -13803 | 17770 | 4.479455 | -3.479455 | 3116 | -13803 | 16919 | 5.429718 | -4.429718 |
| Fam135a | 0.000035 | Gene Expression | mm10 | ENSMUSG00000026153 | True | 4968 | -16780 | 21748 | 4.377617 | -3.377617 | 4516 | -16780 | 21296 | 4.715678 | -3.715678 |
| Tcea1 | 0.000058 | Gene Expression | mm10 | ENSMUSG00000033813 | True | 4251 | -12688 | 16939 | 3.984709 | -2.984709 | 3441 | -12688 | 16129 | 4.687300 | -3.687300 |
| Inpp4a | 0.000072 | Gene Expression | mm10 | ENSMUSG00000026113 | True | 5403 | -15519 | 20922 | 3.872293 | -2.872293 | 4377 | -15519 | 19896 | 4.545579 | -3.545579 |
WARNING: The expression of the highly-expressed gene Rgs20 decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Atp6v1h decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Rb1cc1 decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene St18 decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Pcmtd1 decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Cspp1 decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Arfgef1 decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene A830018L16Rik decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Ncoa2 decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Rpl7 decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Stau2 decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Ube2w decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Gm28376 decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Smap1 decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Fam135a decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Col19a1 decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Lmbrd1 decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Phf3 decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Khdrbs2 decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Bend6 decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Dst decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Fam168b decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Cox5b decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Actr1b decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Tmem131 decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Inpp4a decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene 2010300C02Rik decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Rev1 decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Aff3 decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Npas2 decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Rpl31 decreases quite markedly after CellBender. Check to ensure this makes sense!
WARNING: The expression of the highly-expressed gene Map4k4 decreases quite markedly after CellBender. Check to ensure this makes sense!
Cell probabilities¶
The inferred posterior probability that each droplet is non-empty.
We sometimes write "non-empty" instead of "cell" because dead cells and other cellular debris can still lead to a "non-empty" droplet, which will have a high posterior cell probability. But these kinds of low-quality droplets should be removed during cell QC to retain only high-quality cells for downstream analyses.
Concordance of data before and after remove-background¶
The intent is to change the input data as little as possible while achieving noise removal. These plots show general summary statistics about similarity of the input and output data. We expect to see the data lying close to a straight line (gray). There may be outlier genes/features, which are often those highest-expressed in the ambient RNA.
The plots here show data for inferred cell-containing droplets, and exclude the empty droplets.
PCA of encoded gene expression¶
We are not looking for anything specific in the PCA plot of the gene expression embedding, but often we see clusters that correspond to different cell types. If you see only a single large blob, then the dataset might contain only one cell type, or perhaps there are few counts per droplet.
Summary of warnings:¶
Large deviation in training ELBO from max value late in learning.
Back-tracking in training ELBO.
Expression of gene Rgs20 decreases quite a bit
Expression of gene Atp6v1h decreases quite a bit
Expression of gene Rb1cc1 decreases quite a bit
Expression of gene St18 decreases quite a bit
Expression of gene Pcmtd1 decreases quite a bit
Expression of gene Cspp1 decreases quite a bit
Expression of gene Arfgef1 decreases quite a bit
Expression of gene A830018L16Rik decreases quite a bit
Expression of gene Ncoa2 decreases quite a bit
Expression of gene Rpl7 decreases quite a bit
Expression of gene Stau2 decreases quite a bit
Expression of gene Ube2w decreases quite a bit
Expression of gene Gm28376 decreases quite a bit
Expression of gene Smap1 decreases quite a bit
Expression of gene Fam135a decreases quite a bit
Expression of gene Col19a1 decreases quite a bit
Expression of gene Lmbrd1 decreases quite a bit
Expression of gene Phf3 decreases quite a bit
Expression of gene Khdrbs2 decreases quite a bit
Expression of gene Bend6 decreases quite a bit
Expression of gene Dst decreases quite a bit
Expression of gene Fam168b decreases quite a bit
Expression of gene Cox5b decreases quite a bit
Expression of gene Actr1b decreases quite a bit
Expression of gene Tmem131 decreases quite a bit
Expression of gene Inpp4a decreases quite a bit
Expression of gene 2010300C02Rik decreases quite a bit
Expression of gene Rev1 decreases quite a bit
Expression of gene Aff3 decreases quite a bit
Expression of gene Npas2 decreases quite a bit
Expression of gene Rpl31 decreases quite a bit
Expression of gene Map4k4 decreases quite a bit