1) Make a brand‐new conda env (Python 3.10)

Open your WSL Ubuntu shell and run:

# 1.1) Deactivate any old env
conda deactivate

cd /mnt/c/Users/IAGhe/OneDrive/Documents/Bioinformatics
# 1.2) Create a new env with Python 3.10
conda create -n scvelo-py310 python=3.10 -y

# 1.3) Activate it
conda activate scvelo-py310

# 1.4) Install everything via pip (these versions are known to work together)
pip install numpy pandas scipy matplotlib seaborn scanpy scvelo anndata jupyterlab

---

C.2 Basic scVelo workflow -Toy data (R) Toy Dataset

library(Seurat)
set.seed(123)

# ---- parameters ----
n_genes <- 100
cells_per_cluster <- 25
clusters <- rep(c("A","B"), each = cells_per_cluster)
n_cells <- length(clusters)

# ---- simulate counts ----
# baseline expression ~ Poisson(5)
counts <- matrix(
  rpois(n_genes * n_cells, lambda = 5),
  nrow = n_genes, ncol = n_cells
)
# bump up expression of genes 1:10 in cluster B
counts[1:10, clusters=="B"] <- 
  rpois(sum(clusters=="B") * 10, lambda = 20)

# give the genes & cells names
rownames(counts) <- paste0("Gene", sprintf("%03d", 1:n_genes))
colnames(counts) <- paste0("Cell", sprintf("%02d", 1:n_cells))

# metadata
meta <- data.frame(
  row.names = colnames(counts),
  cluster = clusters
)

# Inspect
head(counts)[,1:6]
#>       Cell01 Cell02 Cell03 Cell04 Cell05 Cell06
#> Gene001      4      8      6      7      3      6
#> Gene002      4      5      7      6      2      5
#> Gene003      5      7      7      6      6      4
#> Gene004      3      6      5      5      4      8
#> Gene005      8      7      1      5      6      7
#> Gene006     10      3      9      2     11      6

head(meta)
#>         cluster
#> Cell01       A
#> Cell02       A
#> Cell03       A
#> Cell04       A
#> Cell05       A
#> Cell06       A

In R, dump your toy matrix and metadata to disk:

write.csv(counts, "counts.csv", quote=FALSE)
write.csv(meta,   "meta.csv",   quote=FALSE)

---

conda activate scvelo-env
conda install -c conda-forge notebook -y

jupyter notebook --no-browser --ip=0.0.0.0 --port=8888

Put in a browser URL:

http://127.0.0.1:8888/tree?token=efe55241d07d4978267af586c6c6e65d7d33aee1e4181eb9

!pip install scikit-misc

Once that finishes, restart your notebook kernel (to pick up the new package).

Then in your notebook run the following cell:

%matplotlib inline
import numpy as np
import pandas as pd
import scanpy as sc
import scvelo as scv

# 0) scVelo / plotting settings
scv.settings.verbosity = 3
scv.settings.set_figure_params('scvelo', figsize=(6,6), dpi=100)

# 1) Load toy data
counts = pd.read_csv("counts.csv", index_col=0)  # 100 genes × 50 cells
meta   = pd.read_csv("meta.csv",   index_col=0)  # 50 cells × 1 column "cluster"

# 2) Build AnnData: cells × genes
adata = sc.AnnData(
    X   = counts.T.values,           # now 50 cells × 100 genes
    obs = meta.copy(),               # must have obs["cluster"]
    var = pd.DataFrame(index=counts.index)
)
adata.obs_names = counts.columns
adata.var_names = counts.index

# 3) Fake velocity layers (unspliced is slightly shifted so the fit isn't degenerate)
np.random.seed(0)
adata.layers["spliced"]   = adata.X.copy()
adata.layers["unspliced"] = adata.X * 0.8 + np.random.normal(
    scale=0.5, size=adata.X.shape
)

# 4) Gene‐level filtering: keep genes seen ≥5× in **both** layers
mask_sp = np.sum(adata.layers["spliced"],   axis=0) >= 5
mask_un = np.sum(adata.layers["unspliced"], axis=0) >= 5
mask    = mask_sp & mask_un
adata   = adata[:, mask]

# 5) HVG selection via Scanpy (top 50 by dispersion)
sc.pp.highly_variable_genes(adata, n_top_genes=50, flavor="seurat_v3")
adata = adata[:, adata.var["highly_variable"]]

# 6) Per‐cell normalization + log:
sc.pp.normalize_total(adata, target_sum=1e4)
sc.pp.log1p(adata)

# 7) PCA / neighbors / UMAP (Scanpy)
sc.pp.pca(adata, n_comps=10, svd_solver="arpack")
sc.pp.neighbors(adata, n_pcs=10, n_neighbors=15)
sc.tl.umap(adata)

# 8) Compute velocity moments (smoothing)
scv.pp.moments(adata, n_pcs=10, n_neighbors=15)

# 9) Fit velocities & graph
scv.tl.velocity(adata, mode="stochastic")
scv.tl.velocity_graph(adata)

# 10) Plot clusters + streams
sc.pl.umap(
    adata,
    color="cluster",
    legend_loc="right margin",
    title="Toy scVelo clusters"
)
scv.pl.velocity_embedding_stream(
    adata,
    basis="umap",
    color="cluster",
    linewidth=1.5,
    legend_loc="right margin",
    title="Toy scVelo velocity"
)

Output:

---

%matplotlib inline

import numpy as np
import pandas as pd
import scanpy as sc
import scvelo as scv

scv.settings.verbosity = 3

scv.settings.set_figure_params('scvelo', figsize=(6,6), dpi=100)

counts = pd.read_csv("counts.csv", index_col=0)

meta = pd.read_csv("meta.csv", index_col=0)

adata = sc.AnnData(
    X   = counts.T.values,
    obs = meta.copy(),
    var = pd.DataFrame(index=counts.index)
)

adata.obs_names = counts.columns

##Output

adata.obs_names = counts.columns

<img width="1588" height="824" alt="image" src="https://github.com/user-attachments/assets/93b63fb7-c6bc-4dda-9495-e5362d9b6a47" />

###Output:

adata.var_names = counts.index

# 3) Fake velocity layers (unspliced is slightly shifted so the fit isn't degenerate)
np.random.seed(0)
adata.layers["spliced"]   = adata.X.copy()
adata.layers["unspliced"] = adata.X * 0.8 + np.random.normal(
    scale=0.5, size=adata.X.shape
)
adata.layers

Output:

np.random.seed(0)

adata.layers["spliced"] = adata.X.copy()

adata.layers["unspliced"] = adata.X * 0.8 + np.random.normal(
    scale=0.5, size=adata.X.shape
)

# 4) Gene‐level filtering: keep genes seen ≥5× in **both** layers
mask_sp = np.sum(adata.layers["spliced"],   axis=0) >= 5
mask_un = np.sum(adata.layers["unspliced"], axis=0) >= 5
mask    = mask_sp & mask_un
adata   = adata[:, mask]

##Output:

mask_sp = np.sum(adata.layers["spliced"], axis=0) >= 5

mask_un = np.sum(adata.layers["unspliced"], axis=0) >= 5

mask = mask_sp & mask_un

adata = adata[:, mask]

sc.pp.highly_variable_genes(adata, n_top_genes=50, flavor="seurat_v3")

adata = adata[:, adata.var["highly_variable"]]

sc.pp.normalize_total(adata, target_sum=1e4)

sc.pp.log1p(adata)

sc.pp.pca(adata, n_comps=10, svd_solver="arpack")

sc.pp.neighbors(adata, n_pcs=10, n_neighbors=15)

sc.tl.umap(adata)

scv.pp.moments(adata, n_pcs=10, n_neighbors=15)

adata.obsp.keys()

adata.uns.keys()

scv.tl.velocity(adata, mode="stochastic")

scv.tl.velocity_graph(adata)

adata.layers["velocity"]

Output:

adata.layers["velocity"].shape

scv.pl.velocity(adata, var_names=adata.var_names[:5], color="cluster")

scv.pl.velocity_embedding_stream(adata, basis="umap", color="cluster")

Final Outputs:

sc.pl.umap(
    adata,
    color="cluster",
    legend_loc="right margin",
    title="Toy scVelo clusters"
)

scv.pl.velocity_embedding_stream(
    adata,
    basis="umap",
    color="cluster",
    linewidth=1.5,
    legend_loc="right margin",
    title="Toy scVelo velocity"
)