Fetal Immune Atlas - Uniforming cell labels¶

In [1]:

import os,sys
import scanpy as sc 
import anndata
import numpy as np
import pandas as pd
import re

In [2]:

%load_ext rpy2.ipython

In [3]:

%%R
library(tidyverse)
library(ggplot2)

R[write to console]: ── Attaching packages ─────────────────────────────────────── tidyverse 1.3.0 ──

R[write to console]: ✔ ggplot2 3.3.2     ✔ purrr   0.3.4
✔ tibble  3.0.4     ✔ dplyr   1.0.2
✔ tidyr   1.1.2     ✔ stringr 1.4.0
✔ readr   1.4.0     ✔ forcats 0.5.0

R[write to console]: ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
✖ dplyr::filter() masks stats::filter()
✖ dplyr::lag()    masks stats::lag()

How are annotations stored?¶

Annotations were saved on the GDrive in Pan_fetal/annotations.

Original files are in original_files. These come from different sources:

fetal_gut_anno.csv was provided by Rasa
fetal_thymus_anno.csv was saved from the obs slot in the anndata object saved as /nfs/team205/ig7/Projects/Pan_fetal/JP_archive/18_Pan_fetal/HTA08.v01.A05.Science_human_fig1.h5ad. I use as labels the most detailed annotation level here (Anno_level_5)
fetal_kidney_anno.csv was saved from the obs slot in the anndata downloaded from https://cellgeni.cog.sanger.ac.uk/BenKidney_v2.1/Fetal_full.h5ad. I use as labels the column celltype
fetal_skin.annotation.20200829.for_Issac.tsv was provided by Ni
All other annotations were given to me by Chenqu as an email attachment on 22/12/2020

The original files are restricted to annotations only in Pan_fetal/annotations/*_annot.csv files.

Uniform labelling¶

As started by Chenqu, we are translating detailed cell labels for each organ to a uniform labelling to use to guide data integration. Translations from old to new labels are saved in Pan_fetal/annotations/*_annot_lvls.csv files.

In [4]:

annot_dir = '/home/jovyan/mount/gdrive/Pan_fetal/annotations/'

Adding translations for missing organs¶

Making uniform labels for thymus cells

In [5]:

th_annot_lvls = {
    "SP T CELL":['CD8+T', 'CD4+T', 'CD8αα(I)', 'αβT(entry)', 'CD8αα(II)', 'Th17', 'γδT', 'CD8+Tmem','CD4+Tmem', 'CD4+CTL', 'Tfh', 'CD4+PD1'],
    "DP T CELL":['DP(P)', 'DP(Q)'],
    "DN T CELL":['DN(Q)', 'DN(P)', "DN(early)"],
    "TREG":['Treg(diff)', 'T(agonist)', "Treg"],
    "MONOCYTE":["Mono"],
    "B CELL":["B_naive", 'B_memory', 'B_plasma'],
    "PRE PRO B CELL":["B_pro/pre"],
    "DC":['DC2', 'DC1',  'aDC1', 'aDC3', 'aDC2'],
    "PDC":['pDC'],
    "NK T": ['NKT'],
    "MACROPHAGE":["Mac"],
    "NEUTROPHIL-MYELOID PROGENITOR":['NMP'],
    "FIBROBLAST":["Fb_1", "Fb_2", "Fb_cycling"],
    "NK":["NK"],
    "MAST CELL":["Mast"],
    "LYMPHOID PROGENITOR":["ETP"],
    "ILC":["ILC3"],
    'ERYTHROCYTE':["Ery"],
    'SMOOTH MUSCLE/PERICYTE':["VSMC"],
    'EPITHELIUM':['mTEC(I)','TEC(neuro)', 'mTEC(II)','cTEC','mcTEC', 'TEC(myo)', 'mTEC(III)','mTEC(IV)', 'Epi_GCM2'],
    'ENDOTHELIAL CELL':["Endo"],
    'MEGAKARYOCYTE':['Mgk'],
    'LYMPHATIC':["Lymph"],
 }

th_lvls_df = pd.DataFrame(pd.concat({k: pd.Series(v) for k, v in th_annot_lvls.items()})).reset_index()
th_lvls_df = th_lvls_df[[0,"level_0"]]
th_lvls_df.columns = ['original_cell_type',"updated_cell_type"]
th_lvls_df.index=th_lvls_df["original_cell_type"]
th_lvls_df = th_lvls_df.drop("original_cell_type",1)
th_lvls_df.to_csv(annot_dir + 'th_annot_lvls.csv')

Making uniform labels for gut cells

In [5]:

gu_annot = pd.read_csv(annot_dir + "gu_annot.csv")

In [6]:

gu_lvls_df = pd.DataFrame(gu_annot["cell.labels"].unique(), columns=['original_cell_type'])

gut_labels = [x.upper() for x in gu_annot["cell.labels"].unique()]
gut_labels = ["ENDOTHELIAL CELL" if "ENDOTHELIAL" in x else x for x in gut_labels]
gut_labels = ["ERYTHROCYTE" if "ERYTHROBLAST" in x else x for x in gut_labels]
gut_labels = ["NEUTROPHIL" if "NEUTROPHIL" in x else x for x in gut_labels]
gut_labels = ["DC" if "DC" in x and x!="PDC" else x for x in gut_labels]
gut_labels = ["FIBROBLAST" if "STROMA" in x else x for x in gut_labels]
gut_labels = ['LYMPHOID PROGENITOR' if x=="CLP" else x for x in gut_labels]
gut_labels = ["ENTEROENDOCRINE CELL" if "EEC" in x else x for x in gut_labels]
gut_labels = ["GLIA" if "GLIA" in x else x for x in gut_labels]
gut_labels = ["NEURON" if "NEURON" in x else x for x in gut_labels]
gut_labels = ["ENDOTHELIAL CELL" if "ENDOTHELIUM" in x else x for x in gut_labels]
gut_labels = ["ENTEROCYTE" if "ENTEROCYTE" in x else x for x in gut_labels]
gut_labels = ["CD16 MYELOID" if "MYELOID" in x else x for x in gut_labels]
gut_labels = ["SP T CELL" if x in ["CYCLING T", 'SELL+ CD4 T', 'SELL+ CD8 T'] else x for x in gut_labels]
gut_labels = ["NEURAL CREST" if "ENCC" in x else x for x in gut_labels]
gut_labels = ["ILC PRECURSOR" if "ILCP" in x else x for x in gut_labels]
gut_labels = ['LTI/ILC3' if 'LTI-LIKE ILC3' in x else x for x in gut_labels]
gut_labels = ['NK T' if 'NK/T' in x else x for x in gut_labels]
gut_labels = ['NK T' if 'NK T CELL' in x else x for x in gut_labels]
gut_labels = ['NK' if 'NK CELL' in x else x for x in gut_labels]
gut_labels = ['MACROPHAGE' if 'MACROPHAGE' in x else x for x in gut_labels]
gut_labels = ['MONOCYTE' if 'MONOCYTE' in x else x for x in gut_labels]
gut_labels = ['MAST CELL' if 'MAST' in x else x for x in gut_labels]
gut_labels = ['SCHWANN CELLS' if 'SCHWANN' in x else x for x in gut_labels]
gut_labels = ['SCHWANN CELLS' if 'SCP' in x else x for x in gut_labels]
gut_labels = ['B CELL' if 'IMMATURE B' in x else x for x in gut_labels]
gut_labels = ['PRE B CELL' if 'PRE-B' in x else x for x in gut_labels]
gut_labels = ['PRO B CELL' if 'PRO-B' in x else x for x in gut_labels]
gut_labels = ['SMOOTH MUSCLE/PERICYTE' if 'SMC' in x else x for x in gut_labels]

gu_lvls_df["updated_cell_type"] = gut_labels

gu_lvls_df.index = gu_lvls_df["original_cell_type"]
gu_lvls_df = gu_lvls_df.drop("original_cell_type",1)
gu_lvls_df.to_csv(annot_dir + 'gu_annot_lvls.csv')

Making uniform labels for kidney cells: I use the compartment annotation from the original file to group nephron annotations

N.B. in the original file there was a grand total of 12 cells with duplicated barcodes: I decided to exclude these

In [8]:

ki_annot = pd.read_csv(annot_dir + "ki_annot.csv")

In [9]:

ki_lvls_df = pd.DataFrame(ki_annot["cell.labels"].unique(), columns=['original_cell_type'])

ki_labels = [x.upper() for x in ki_annot["cell.labels"].unique()]

## Pick kidney specific labels
ki_anno_df = pd.read_csv(annot_dir + "original_files/fetal_kidney_anno.csv", index_col=0)
ki_anno_df = ki_anno_df[["celltype", "compartment"]].drop_duplicates()
ki_anno_df.celltype = [x.upper() for x in ki_anno_df.celltype]
fetal_nephron_labels = ki_anno_df[ki_anno_df.compartment=="fetal_nephron"]["celltype"].values

ki_labels = ['MACROPHAGE' if 'MACROPHAGE' in x else x for x in ki_labels]
ki_labels = ["SP T CELL" if " T CELL" in x else x for x in ki_labels]
ki_labels = ["NK" if "NK" in x else x for x in ki_labels]
ki_labels = ["B CELL" if "B CELL" in x else x for x in ki_labels]
ki_labels = ['FIBROBLAST' if x in ['PROLIFERATING FIBROBLAST', 'FIBROBLAST 1', 'FIBROBLAST 2'] else x for x in ki_labels]
ki_labels = ['MYOFIBROBLAST' if x in ['PROLIFERATING MYOFIBROBLAST', 'MYOFIBROBLAST 1', 'MYOFIBROBLAST 2'] else x for x in ki_labels]
ki_labels = ['DC' if "CDC" in x else x for x in ki_labels]
ki_labels = ['PDC' if "PDC" in x else x for x in ki_labels]
ki_labels = ['MONOCYTE' if "MONOCYTE" in x else x for x in ki_labels]
ki_labels = ['STROMA PROGENITOR' if "STROMA PROGENITOR" in x else x for x in ki_labels]
ki_labels = ['MAST CELL' if "MAST" in x else x for x in ki_labels]
ki_labels = ['ILC' if x=="INNATE LIKE LYMPHOCYTE" in x else x for x in ki_labels]
ki_labels = ['ENDOTHELIAL CELL' if "ENDOTHELIUM" in x else x for x in ki_labels]
ki_labels = ['NEPHRON' if x in fetal_nephron_labels else x for x in ki_labels]

ki_lvls_df["updated_cell_type"] = ki_labels

ki_lvls_df.index = ki_lvls_df["original_cell_type"]
ki_lvls_df = ki_lvls_df.drop("original_cell_type",1)
ki_lvls_df.to_csv(annot_dir + 'ki_annot_lvls.csv')

Updated from skin

In [111]:

sk_annot = pd.read_table(annot_dir + "original_files/fetal_skin.annotation.20200829.for_Issac.tsv", header=None)

In [112]:

sk_labels = [x.upper() for x in sk_annot.iloc[:,1].unique()]

sk_lvls_df = pd.read_csv(annot_dir + "original_files/sk_annot_lvls.csv")
sk_lvls_df.iloc[:,1] = [x.upper() for x in sk_lvls_df.iloc[:,1]]
sk_lvls_df.iloc[:,0] = [x.upper() for x in sk_lvls_df.iloc[:,0]]

sk_labels = [sk_lvls_df.updated_cell_type[sk_lvls_df.original_cell_type==x] if x in sk_lvls_df.original_cell_type else x for x in sk_labels]

sk_lvls_df = pd.DataFrame(sk_annot.iloc[:, 1].unique(), columns=['original_cell_type'])

sk_labels = ['GMP' if 'GRANULOCYTE PROGENITOR' in x else x for x in sk_labels]
sk_labels = ["GLIA" if "GLIA" in x else x for x in sk_labels]
sk_labels = ["FIBROBLAST" if "FIBROBLAST" in x else x for x in sk_labels]
sk_labels = ["ENDOTHELIAL CELL" if "ENDOTHELIUM" in x else x for x in sk_labels]
sk_labels = ["MONOCYTE/DC PRECURSOR" if "MONOCYTE PRECUSOR" in x else x for x in sk_labels]
sk_labels = ["NK" if x=="NK CELL" else x for x in sk_labels]
sk_labels = ["NK T" if x=="NK T CELL" else x for x in sk_labels]
sk_labels = ["SMOOTH MUSCLE/PERICYTE" if x=="SMOOTH MUSCLE" else x for x in sk_labels]
sk_labels = ["MUSCLE" if x=="SKELETAL MUSCLE" else x for x in sk_labels]
sk_labels = ["LTI/ILC3" if x in ["LTI", "ILC3"] else x for x in sk_labels]
sk_labels = ["MONOCYTE/MACROPHAGE" if x in ["MAC-MONO"] else x for x in sk_labels]
sk_labels = ["LANGERHAN CELLS" if x in ['LANGERHANS CELL'] else x for x in sk_labels]
sk_labels = ["NEURON" if x in ['NEURONAL CELLS'] else x for x in sk_labels]
sk_labels = ["SCHWANN CELLS" if x in ['SCHWANN CELL'] else x for x in sk_labels]
sk_labels = ["DC" if x in ['ASDC'] else x for x in sk_labels]
sk_labels = ["KERATINOCYTES" if x in ['SUPRABASAL KC', "HF KC"] else x for x in sk_labels]

sk_lvls_df["updated_cell_type"] = sk_labels

sk_lvls_df.index = sk_lvls_df["original_cell_type"]
sk_lvls_df = sk_lvls_df.drop("original_cell_type",1)
sk_lvls_df.to_csv(annot_dir + 'sk_annot_lvls.csv')

In [113]:

sk_lvls_df

Out[113]:

	updated_cell_type
original_cell_type
Macrophage	MACROPHAGE
Monocyte	MONOCYTE
Granulocyte progenitor	GMP
DC2	DC2
Microglia	GLIA
Monocyte precusor	MONOCYTE/DC PRECURSOR
Fibroblast WNT2+	FIBROBLAST
NK cell	NK
B cell	B CELL
Lti	LTI/ILC3
ILC3	LTI/ILC3
MEMP	MEMP
CD4 T cell	CD4 T CELL
HSC	HSC
Mac-mono	MONOCYTE/MACROPHAGE
Lymphatic endothelium	ENDOTHELIAL CELL
NK T cell	NK T
CD8 T cell	CD8 T CELL
Skeletal muscle	MUSCLE
pDC	PDC
Mast cell	MAST CELL
DC1	DC1
Vascular endothelium	ENDOTHELIAL CELL
Neutrophil	NEUTROPHIL
Langerhans cell	LANGERHAN CELLS
Megakaryocyte	MEGAKARYOCYTE
Erythroid	ERYTHROID
Myofibroblast	FIBROBLAST
Fibroblast HF	FIBROBLAST
Fibroblast HOX5+	FIBROBLAST
Fibroblast unknown	FIBROBLAST
Pericyte	PERICYTE
Fibroblast FRZB+	FIBROBLAST
Early erythroid	EARLY ERYTHROID
Neuronal cells	NEURON
Schwann cell	SCHWANN CELLS
Smooth muscle	SMOOTH MUSCLE/PERICYTE
ILC2	ILC2
Lymphoid progenitor	LYMPHOID PROGENITOR
Adipocyte	ADIPOCYTE
Treg	TREG
ASDC	DC
Periderm	PERIDERM
Melanocyte	MELANOCYTE
Melanoblast	MELANOBLAST
Basal KC	BASAL KC
Suprabasal KC	KERATINOCYTES
Fibroblast dermal papillia	FIBROBLAST
HF KC	KERATINOCYTES

Make uniform labelling¶

In [114]:

annot_files = [x for x in os.listdir(annot_dir) if "annot.csv" in x]
lvls_annot_files = [x for x in os.listdir(annot_dir) if "annot_lvls" in x]

organs = [x.split("_annot")[0] for x in annot_files]

Load all translations to fix errors from manual annotation

In [115]:

lvls_labels_full = pd.DataFrame()
for file in lvls_annot_files:
    ## Read labels file
    organ = file.split("_")[0]
    lvls_labels_df = pd.read_csv(annot_dir + file)
    lvls_labels_df["organ"] = organ
    lvls_labels_full = pd.concat([lvls_labels_full, lvls_labels_df], 0)

Some notes:

convert CD4 and CD8 to T cell
Are 'muscle' and 'muscle/myofibroblast' distinct?
Are precursor and progenitor the same?
Are "pre pro B" and "Pre to pro B" the same?
not sure where to place Th17 (I put it in SP according to JPs broad annotation)
I put ETP from thymus with Early erythroid T cell, is that correct?
'ENDOTHELIAL CELL' and 'MSC/ENDOTHELIAL CELL'?
cycling myeloid with CD16+?
stromal cells in gut?
uniforming neural crest
erythroblasts outside of the gut?

In [116]:

lvls_labels_full["uniform_label"] = [x.upper() for x in lvls_labels_full.updated_cell_type]

## Fix some typos
lvls_labels_full["uniform_label"] = ['KUPFFER CELL' if x=='KUPFFER ELL' else x for x in lvls_labels_full["uniform_label"]]
lvls_labels_full["uniform_label"] = ['ENDOTHELIAL CELL' if x=='ENDOTHELIUM' else x for x in lvls_labels_full["uniform_label"]]
lvls_labels_full["uniform_label"] = ['EO/BASO/MAST PRECURSOR' if x=='EO/BASE/MAST PRECURSOR' else x for x in lvls_labels_full["uniform_label"]]
lvls_labels_full["uniform_label"] = ['MAST CELL' if x=='MAST CEL' else x for x in lvls_labels_full["uniform_label"]]
lvls_labels_full["uniform_label"] = ['NEUTROPHIL' if x=='NEUTROPHILE' else x for x in lvls_labels_full["uniform_label"]]
lvls_labels_full["uniform_label"] = ['NEUTROPHIL-MYELOID PROGENITOR' if x in ['NEUT-MYELOID PRECURSOR', 'NEUTROPHIL MYELOID PROGENITOR'] else x for x in lvls_labels_full["uniform_label"]]
lvls_labels_full["uniform_label"] = ['PRE PRO B CELL' if x=='PRE PRO B' else x for x in lvls_labels_full["uniform_label"]]
# lvls_labels_full["uniform_label"] = ['PRE PRO B CELL' if x=='PRE TO PRO B CELL' else x for x in lvls_labels_full["uniform_label"]]
lvls_labels_full["uniform_label"] = ['PRO B CELL' if x=='PRO B' else x for x in lvls_labels_full["uniform_label"]]
lvls_labels_full["uniform_label"] = ['PRE B CELL' if x=='PRE B' else x for x in lvls_labels_full["uniform_label"]]
lvls_labels_full["uniform_label"] = ['NEURAL CREST' if 'NEURAL CREST' in x else x for x in lvls_labels_full["uniform_label"]]
lvls_labels_full["uniform_label"] = ['GLIA' if 'GLIA' in x else x for x in lvls_labels_full["uniform_label"]]
lvls_labels_full["uniform_label"] = ['SP T CELL' if x in ["CD8 T CELL", "CD4 T CELL", "T CELL"] else x for x in lvls_labels_full["uniform_label"]]
lvls_labels_full["uniform_label"] = ['ERYTHROID' if x in ['ERYTHROID','YS ERYTHROID'] else x for x in lvls_labels_full["uniform_label"]]
lvls_labels_full["uniform_label"] = ['DC' if x in ['DC1','DC2'] else x for x in lvls_labels_full["uniform_label"]]
lvls_labels_full["uniform_label"] = ['ILC' if x=='ILC2' else x for x in lvls_labels_full["uniform_label"]]
lvls_labels_full["uniform_label"] = ['ILC' if x=='ILC2' else x for x in lvls_labels_full["uniform_label"]]

lvls_labels_full.reset_index(inplace=True)

In [117]:

## Save uniform labelling 
lvls_labels_full[['original_cell_type', "organ", "uniform_label"]].to_csv(annot_dir + 'uniform_lvls.csv')

In [10]:

lvls_labels_full = pd.read_csv(annot_dir + 'uniform_lvls.csv')

In [13]:

lvls_labels_full[lvls_labels_full.uniform_label_expanded_merged=="ERYTHROID"]

Out[13]:

	Unnamed: 0	original_cell_type	organ	uniform_label	uniform_label_lvl0	uniform_label_expanded_merged
25	25	immature EC	bm	ERYTHROCYTE	ERYTHROID	ERYTHROID
59	59	sinusoidal EC	bm	ERYTHROCYTE	ERYTHROID	ERYTHROID
62	62	tip EC	bm	ERYTHROCYTE	ERYTHROID	ERYTHROID
79	79	Enterocyte	gu	ENTEROCYTE	EPI	ERYTHROID
83	83	Erythroblast	gu	ERYTHROCYTE	ERYTHROID	ERYTHROID
96	96	BEST4 enterocyte	gu	ENTEROCYTE	EPI	ERYTHROID
165	165	Erythroid	ki	ERYTHROID	ERYTHROID	ERYTHROID
241	241	Erythroid	sk	ERYTHROID	ERYTHROID	ERYTHROID
270	270	Erythroid	sp	ERYTHROID	ERYTHROID	ERYTHROID
351	351	Ery	th	ERYTHROCYTE	ERYTHROID	ERYTHROID
372	372	Erythroid	ys	ERYTHROID	ERYTHROID	ERYTHROID

In [118]:

%%R -i lvls_labels_full -w 1200 -h 900

lvls_labels_full %>%
arrange(uniform_label) %>%
mutate(original_cell_type=factor(original_cell_type, levels=unique(original_cell_type))) %>%
ggplot(aes(original_cell_type, uniform_label, fill=organ)) +
geom_tile() +
theme_bw(base_size=12) +
theme(axis.text.x=element_text(angle=90, hjust=1))

Add uniform labels to full cell annotation

In [119]:

cell_labels_full = pd.DataFrame()
for organ,file in zip(organs, annot_files):
    ## Read labels file
    orig_labels_df = pd.read_csv(annot_dir + file,  index_col=0)
    orig_labels_df["organ"] = organ
    
    if os.path.exists(annot_dir + organ + "_annot_lvls.csv"):
        ## Read labels translations
        new_labels = lvls_labels_full[lvls_labels_full.organ==organ]
        new_labels.index = new_labels["original_cell_type"]
        ## Add updated labels
        orig_labels_df['uniform_label'] = new_labels.loc[orig_labels_df["cell.labels"]]['uniform_label'].values
    
    cell_labels_full = pd.concat([cell_labels_full, orig_labels_df], 0)

In [120]:

cell_labels_full.to_csv(annot_dir + "uniform_labels_full.csv")

In [121]:

labels_count = cell_labels_full.reset_index()[["uniform_label", "organ", "index"]].groupby(["uniform_label", "organ"]).count().reset_index()
labels_count.columns = ["uniform_label", "organ", "n_cells"]

In [122]:

%%R -i labels_count -h 1000 -w 500
library(ggplot2)
library(dplyr)

labels_count %>%
ggplot(aes(uniform_label, n_cells, fill=organ)) +
geom_col() +
scale_y_log10() +
coord_flip()

Add manually annotated level_0¶

In [133]:

lvl0 = pd.read_csv("/home/jovyan/mount/gdrive/Pan_fetal/annotations/lvl_0_unified_labels/uniform_labels_lvl0_20210118.csv", index_col=0)
lvl0.index = lvl0.uniform_label

# lvls_labels_full.drop("index",1, inplace=True)
lvls_labels_full["uniform_label_lvl0"] = lvl0.loc[lvls_labels_full.uniform_label]["uniform_label_lvl0"].values
lvls_labels_full["uniform_label_expanded_merged"] = lvl0.loc[lvls_labels_full.uniform_label]["uniform_label_expanded_merged"].values

In [134]:

lvls_labels_full

Out[134]:

	original_cell_type	updated_cell_type	organ	uniform_label	uniform_label_lvl0	uniform_label_expanded_merged
0	adipo-CAR	Adipocyte	bm	ADIPOCYTE	ADIPOSE	ADIPOCYTE
1	arteriolar fibroblast	Fibroblast	bm	FIBROBLAST	STROMAL	FIBROBLAST
2	basophil	Basophil	bm	BASOPHIL	EO_BASO_MAST	BASOPHIL
3	CD14 monocyte	Monocyte	bm	MONOCYTE	MYELOID	MONOCYTE
4	CD4 T cell	CD4 T cell	bm	SP T CELL	LYMPHOID	SP T CELL
...	...	...	...	...	...	...
390	MPP2	MPP	ys	MPP	HSC_MPP	MPP
391	Neutrophil_precursor	Neutrophil precursor	ys	NEUTROPHIL PRECURSOR	MYELOID	NEUTROPHIL PRECURSOR
392	NK	NK	ys	NK	LYMPHOID	NK
393	Promonocyte	Promonocyte	ys	PROMONOCYTE	MYELOID	PROMONOCYTE
394	Smooth_Muscle	Muscle	ys	MUSCLE	MUSCLE	MUSCLE

395 rows × 6 columns

In [135]:

## Save uniform labelling 
lvls_labels_full[['original_cell_type', "organ", "uniform_label", "uniform_label_lvl0", "uniform_label_expanded_merged"]].to_csv(annot_dir + 'uniform_lvls.csv')

In [136]:

lvl0_matching = lvls_labels_full[['uniform_label', "uniform_label_lvl0", "uniform_label_expanded_merged"]].drop_duplicates()
lvl0_matching.index = lvl0_matching["uniform_label"] 
lvl0_full = lvl0_matching.loc[cell_labels_full["uniform_label"]][["uniform_label_lvl0", "uniform_label_expanded_merged"]]

In [137]:

cell_labels_full["uniform_label_lvl0"] = lvl0_full["uniform_label_lvl0"].values
cell_labels_full["uniform_label_expanded_merged"] = lvl0_full["uniform_label_expanded_merged"].values

In [138]:

cell_labels_full.to_csv(annot_dir + "uniform_labels_full.csv")