Create app.py

app.py

@@ -0,0 +1,139 @@
+import datasets
+import streamlit as st
+import numpy as np
+import pandas as pd
+import altair as alt
+
+st.set_page_config(layout='wide')
+
+# parse out feature_names from URL query args to it's possible to link to this page
+query_params = st.query_params
+if "feature_name_1" in query_params.keys():
+    feature_name_1 = query_params["feature_name_1"]
+else:
+    feature_name_1 = "C3_06330W_A"
+
+if "feature_name_2" in query_params.keys():
+    feature_name_2 = query_params["feature_name_2"]
+else:
+    feature_name_2 = "C3_06180C_A"
+
+    
+
+
+st.markdown("""
+# CaurisCEN Expression Scatter
+**CaurisCEN** is a co-expression network for *Candida auris* built on 577 RNA-seq runs across 2 96-well plates formats in 3 biological replicas.
+A pair of genes are said to be co-expressed when their expression is correlated across different conditions and
+is often a marker for genes to be involved in similar processes. 
+To Cite:
+    Rapala JR, MJ O'Meara, TR O'Meara
+    CaurisCEN: A Co-Expression Network for Candida auris
+* Code available at https://github.com/maomlab/CalCEN/tree/master/vignettes/CaurisCEN
+* Full network and dataset: https://huggingface.co/datasets/maomlab/CaurisCEN
+
+## Plot scatter plot expression for a pair of genes across studies.
+Put in the ``B9J08  e.g. B9J08_000884`` chromosome feature name for two genes.
+""")
+
+gene_metadata = datasets.load_dataset(
+    path = "maomlab/CaurisCEN",
+    name = "gene_metadata",
+    data_dir = "gene_metadata/data")['train'].to_pandas()
+
+expression_runs = datasets.load_dataset(
+    path = "maomlab/CaurisCEN",
+    name = "run_metadata",
+    data_dir = "run_metadata/data")['train'].to_pandas()
+
+expression_matrix = datasets.load_dataset(
+    path = "maomlab/CaurisCEN",
+    name = "expression_matrix",
+    data_dir = "expression_matrix")['train'].to_pandas()
+
+
+#DEBUG
+print(f"estimated_expression shape: {estimated_expression.shape}")
+
+col1, col2, col3, padding = st.columns(spec = [0.2, 0.2, 0.2, 0.4])
+with col1:
+    feature_name_1 = st.text_input(
+        label = "Feature Name 1",
+        value = f"{feature_name_1}",
+        max_chars = 10,
+        help = "B9J08 Locus Tag e.g. B9J08_000884")
+
+with col2:
+    feature_name_2 = st.text_input(
+        label = "Feature Name 2",
+        value = f"{feature_name_2}",
+        max_chars = 10,
+        help = "B9J08 New Locus Tag e.g. B9J08_004112")
+
+# check the user input
+try:
+    feature_name_1 = gene_metadata.loc[gene_metadata["locus_tag_old"] == feature_name_1]["feature_name"].values[0]
+    gene_name_1 = gene_metadata.loc[gene_metadata["locus_tag_old"] == feature_name_1]["gene_name"].values[0]
+    description_1 = gene_metadata.loc[gene_metadata["locus_tag_old"] == feature_name_1]["description"].values[0]
+except:
+    st.error(f"Unable to locate Feature Name for gene 1: {feature_name_1}, it should be of the form 'B9J08_######'")
+
+try:
+    feature_name_2 = gene_metadata.loc[chromosome_features["locus_tag_old"] == feature_name_2]["feature_name"].values[0]
+    gene_name_2 = gene_metadata.loc[chromosome_features["locus_tag_old"] == feature_name_2]["gene_name"].values[0]
+    description_2 = gene_metadata.loc[chromosome_features["locus_tag_old"] == feature_name_2]["description"].values[0]
+except:
+    st.error(f"Unable to locate Feature Name for gene 2: {feature_name_2}, it should be of the form 'B9J08_######'")
+
+expression_1 = expression_matrix.loc[expression_matrix["gene"] == feature_name_1].transpose()[1:].to_numpy()
+expression_2 = expression_matrix.loc[expression_matrix["gene"] == feature_name_2].transpose()[1:].to_numpy()
+
+chart_data = pd.DataFrame({
+    "feature_name_1": feature_name_1,
+    "feature_name_2": feature_name_2,
+    "expression_1": expression_1,
+    "expression_2": expression_2,
+    "log_expression_1": np.log10(expression_1 + 1),
+    "log_expression_2": np.log10(expression_2 + 1),
+    "run_accession": expression_matrix.columns[1:]})
+chart_data = chart_data.merge(
+    right = expression_runs,
+    on = "run_accession")
+
+with col3:
+    st.text('') # help alignment with input box
+    st.download_button(
+        label = "Download data as TSV",
+        data = chart_data.to_csv(sep ='\t').encode('utf-8'),
+        file_name = f"CalCEN_expression_{feature_name_1}_vs_{feature_name_2}.tsv",
+        mime = "text/csv")
+
+
+st.markdown(f"""
+#### Gene 1:
+* *Gene ID*: [{feature_name_1}](http://www.candidagenome.org/cgi-bin/locus.pl?locus={feature_name_1}&organism=C_auris_B8441)
+{'* *Gene Name*:' + gene_name_1 if gene_name_1 is not None else ''}
+* *Description*: {description_1}
+* *Top [Co-Expressed Partners](https://huggingface.co/spaces/maomlab/CaurisCEN-TopHits?feature_name={feature_name_1})*
+
+#### Gene 2:
+* *Gene ID*: [{feature_name_2}](http://www.candidagenome.org/cgi-bin/locus.pl?locus={feature_name_2}&organism=C_auris_B8441)
+{'* *Gene Name*:' + gene_name_2 if gene_name_2 is not None else ''}
+* *Description*: {description_2}
+* *Top [Co-Expressed Partners](https://huggingface.co/spaces/maomlab/CaurisCEN-TopHits?feature_name={feature_name_2})*
+""")
+
+chart = (
+   alt.Chart(
+       chart_data,
+       width = 750,
+       height = 750)
+   .mark_circle()
+   .encode(
+       x=alt.X("log_expression_1", title=f"Log10[{feature_name_1}+1] Expression"),
+       y=alt.Y("log_expression_2", title=f"Log10[{feature_name_2}+1] Expression"),
+       color=alt.Color("study_accession", title="Study Accession"),
+       tooltip=["run_accession", "study_accession"]))
+
+st.altair_chart(
+    chart)