In VPetukhov/dropEstAnalysis: Analysis to produce figures from dropEst paper

Content by themes

1. Dataset annotations:
   1. 10x Frozen BMMCs (Healthy Control 1)
   2. 10x 8k PBMCs from a Healthy Donor
   3. inDrop Mouse Bone Marrow cells
2. Origin of background cells in Human/mouse datasets:
   1. Drop-seq thousand
   2. 10x 1k 1:1 Mixture
   3. 10x 6k 1:1 Mixture
3. Filtration of low-quality cells
   1. Robustness of classifier to noise
   2. Validation of inDrop Mouse BMCs
   3. Validation of inDrop Mouse PCs
   4. Validation of 10x Human BMMCs
   5. Validation of 10x 8k PBMCs
4. UMI corrections:
   1. Correction effect on 10x BMMCs dataset
   2. UMI trimming on 10x AML035 Post-transplant dataset

5. Cell barcode merge validation

   1. Merge of human/mouse mixtures
   2. Number of molecules per cell

6. Runtimes

   1. inDrop BMCs
   2. inDrop ESCs

Content by figures

Some figures were created with Python code and they are not published here. Please, write personally to request the code.

Main figures

1. Figure 1. Skewed distribution of UMIs leads to increased number of UMI collisions.
   • ~~A. Skewness of UMI distribution~~
   • ~~B. Uneven nucleotide frequencies~~
   • C. Estimated number of collisions
2. Figure 2. Comparison of UMI collision and sequencing error correction methods.
   • A. Collisions on trimmed UMIs
   • B. Collision adjustment for trimmed UMIs
   • C. Correction of UMI errors on trimmed data
   • D. Distribution of edit distances between pairs of UMIs
   • E. Magnitude of correction
3. Figure 3. Correcting for Cellular Barcode errors.
   • A. Origin of background cells
   • ~~B. Number of equidistant adjacent CBs~~
   • C. Increase in #molecules per cell
4. Figure 4. Selection of the optimal size threshold for 10x BMMCs dataset.
   • A. The distribution of molecular mass across CBs of different sizes
   • B. Gene expression profiles of cluster-specific genes
   • C. Gene expression profiles of cluster-specific genes
   • D. t-SNE visualization
5. Figure 5. Filtration of low-quality cells for the 10x 8k PBMCs dataset.
   • A. t-SNE visualization
   • B. Cells with high mitochondrial fraction
   • C. Cells with high intergenic fraction
   • D. Cells with low #UMIs per gene
6. Figure 6. Filtration of low-quality cells for the inDrop mouse BMCs dataset.
   • A. Gene expression profiles of cluster-specific genes
   • B. Gene expression profiles of cluster-specific genes
   • C. t-SNE visualization

Supplementary figures

1. ~~S1. Skewness of UMI distributions.~~
2. S2. Simulation of UMI collision frequencies
3. ~~S3. Probability of observing adjacent UMIs in small genes.~~
4. ~~S4. Recognition of UMI errors by base calling quality.~~
5. S5. Impact of non-uniform distribution on UMI collisions
6. S6. UMI collisions on trimmed data
7. S7. Magnitude of UMI correction
8. S8. Comparison of UMI correction algorithms on trimmed data
9. S9. Initial labeling of high-quality cells based on cell size distributions
10. S10. Human and mouse cell mixture dataset by 10x
11. S11. Robustness of different classifiers to training errors
12. S12. Annotation of the 10x Frozen BMMCs dataset
13. S13. Annotation of the 10x 8k PBMCs dataset
14. S14. Annotation of the inDrop BMCs dataset
15. S15. Classification of low- and high-quality cells on 10x data
16. S16. Comparison of the initial label assignments with the cell quality score predicted by the algorithm
17. S17. Classification of low- and high-quality cells on inDrop mouse pancreatic cells data

Content by tables

Main tables

1. Table 1. Analysis of merge targets on human/mouse mixture datasets
   • 10x
   • Drop-seq
2. Table 2. 5-fold CV results (mean ± sd)

Supplementary tables

1. Table S1. Gene markers, used for annotation of the 10x Frozen BMMCs
2. Table S2. Gene markers, used for annotation of the 10x 8k PBMCs dataset
3. Table S3. Gene markers, used for annotation of the inDrop mouse BMCs dataset
4. Table S4. Fraction of rescued cells for the 10x Frozen BMMCs dataset
5. Table S5. Fraction of rescued cells for the 10x 8k PMMCs dataset
6. Table S6. Fraction of rescued cells for the inDrop mouse PCs dataset
7. Table S7. Fraction of rescued cells for the inDrop mouse BMCs dataset
8. Table S8. Runtimes of dropEst pipeline
   • inDrop Mouse BMCs
   • inDrop Mouse ESCs

VPetukhov/dropEstAnalysis documentation built on Dec. 28, 2019, 8:16 p.m.