Hey everyone,
Trying to use the mouse classifier on Garnett for our rat / human tissue data:
colData(wt_rn)$garnett_cluster <- clusters(wt_rn)
mouse_wt_rn_classifier <- train_cell_classifier(cds=wt_rn,marker_file =
"./mmBrain_markers.txt",
db=org.Rn.eg.db::org.Rn.eg.db,
cds_gene_id_type = "ENSEMBL",
num_unknown = 50,
marker_file_gene_id_type = "SYMBOL",
cores=4)
For some reason, when I remove cores=4 (I've run my own training classifier on cores=4 without fail), it classifies, but when I add it it throws:
<simpleError in file(con, "w"): all connections are in use>
GLMNET failed
---
When removing cores = 4, it seems to classify:
training_sample
CNS glia Immune cells Neural crest-like glia Neurons Unknown Vascular cells
500 500 500 500 20 279
Model training finished.
training_sample
Cerebellum neurons Cholinergic monoaminergic and peptidergic neurons
26 18
Di- and mesencephalon neurons Enteric neurons
111 500
Hindbrain neurons Immature neural
65 14
Peripheral sensory neurons Spinal cord neurons
51 11
Telencephalon interneurons Telencephalon projecting neurons
114 112
Unknown
20
Model training finished.
training_sample
Microglia Perivascular macrophages Unknown
500 500 20
Model training finished.
training_sample
Pericytes Unknown Vascular leptomeningeal cells Vascular smooth muscle cells
134 20 40 228
Model training finished.
training_sample
Astroependymal cells Oligodendrocytes Unknown
236 500 20
Model training finished.
training_sample
Enteric glia Enteric mesothelial fibroblasts Olfactory ensheathing cells Oligodendrocytes precursor cells
500 2 500 19
Satellite glia Unknown
418 20
The following cell types do not have enough training cells and will be dropped: Enteric mesothelial fibroblasts
Model training finished.
training_sample
Granular layer interneurons cerebellum Molecular layer interneurons cerebellum Neuroblasts cerebellum
4 104 8
Unknown
20
The following cell types do not have enough training cells and will be dropped: Granular layer interneurons cerebellum
Model training finished.
training_sample
Cholinergic and monoaminergic neurons Peptidergic neurons Unknown
500 44 20
Model training finished.
training_sample
Di- and mesencephalon excitatory neurons Di- and mesencephalon inhibitory neurons Unknown
367 90 20
Model training finished.
training_sample
Excitatory neurons hindbrain hind Inhibitory neurons spinal cord Unknown
69 500 20
Model training finished.
training_sample
Spinal cord excitatory neurons Unknown
164 20
Model training finished.
training_sample
Non-glutamatergic neuroblasts Unknown
95 20
Model training finished.
training_sample
Olfactory inhibitory neurons Telencephalon inhibitory interneurons Unknown
342 500 20
Model training finished.
training_sample
Dentate gyrus granule neurons Telencephalon projecting excitatory neurons Telencephalon projecting inhibitory neurons
41 204 500
Unknown
20
Model training finished.
training_sample
Cholinergic enteric neurons Nitrergic enteric neurons Unknown
500 2 20
The following cell types do not have enough training cells and will be dropped: Nitrergic enteric neurons
Model training finished.
training_sample
Peripheral sensory neurofilament neurons Peripheral sensory non-peptidergic neurons Peripheral sensory peptidergic neurons
315 19 500
Unknown
20
Model training finished.
No cells classified as Sympathetic neurons. No subclassification
training_sample
Astrocytes Choroid epithelial cells Ependymal cells
500 500 348
Subcommissural organ hypendymal cells Unknown
2 20
The following cell types do not have enough training cells and will be dropped: Subcommissural organ hypendymal cells
Model training finished.
training_sample
Committed oligodendrocytes cells Mature oligodendrocytes Myelin forming oligodendrocytes Newly formed oligodendrocytes
180 500 115 500
Unknown
20
Model training finished.
training_sample
Adrenergic cell groups of the medulla Afferent nuclei of cranial nerves VI-XII
216 14
Cholinergic neurons septal nucleus Meissnert and diagonal band Dopaminergic neurons periaqueductal grey
501 47
Glutamatergic projection neurons of the raphe nucleus Noradrenergic neurons of the medulla
11 24
Orexin-producing neurons hypothalamus Pmch neurons hypothalamus
3 4
Serotonergic neurons Unknown
500 20
The following cell types do not have enough training cells and will be dropped: Orexin-producing neurons hypothalamus Pmch neurons hypothalamus
Model training finished.
training_sample
Central canal neurons spinal cord Inhibitory neurons septal nucleus Inhibitory neurons telencephalon
6 31 120
Oxytocin-producing cells hypothalamus Peptidergic neurons hypothalamus Unknown
6 36 20
The following cell types do not have enough training cells and will be dropped: Central canal neurons spinal cord Oxytocin-producing cells hypothalamus
Model training finished.
training_sample
Excitatory neurons hindbrain Excitatory neurons hypothalamus Excitatory neurons midbrain Excitatory neurons thalamus
44 500 50 33
Unknown
20
Model training finished.
training_sample
Inhibitory neurons hindbrain di Inhibitory neurons hypothalamus di Inhibitory neurons midbrain di Unknown
6 126 500 20
---
After it does the training, when going to classify cells:
wt_rn <- classify_cells(wt_rn, mouse_classifier,
db = org.Rn.eg.db,
cluster_extend = TRUE,
cds_gene_id_type = "ENSEMBL")
All of the cells are marked as "Unknown" rather than the cells it seems to have classified using the mouse marker files from Garnett.
Is it just there is no way to adapt the classifiers across-species? Or am I missing something?
Thanks!
James