cell_signaling / app.py
ProfRick's picture
Update app.py
eca077f verified
# app.py
import streamlit as st
import numpy as np
import matplotlib.pyplot as plt
import random
st.set_page_config(page_title="Cell–Cell Communication Builder", layout="wide")
# -----------------------------
# Base option sets (your terms)
# -----------------------------
SECRETING_CELLS_BASE = [
"— select —",
"Presynaptic neuron",
"Hypothalamus/Pituitary/Adrenal cortex",
"Cardiomyocyte",
"Tumor cell",
]
MOLECULES_BASE = [
"— select —",
"EGF",
"Neurotransmitter",
"Intracellular ions",
"Cortisol",
]
RECEIVING_CELLS_BASE = [
"— select —",
"Neighboring cardiomyocytes",
"Same cell",
"Postsynaptic cell",
"Liver & skeletal muscle",
]
SIGNAL_TYPES_BASE = [
"— select —",
"direct",
"autocrine",
"paracrine",
"endocrine",
]
MOLECULE_CLASS_BASE = [
"— select —",
"hydrophilic",
"lipophilic",
"N/A (not applicable)",
]
RECEPTOR_LOCS_BASE = [
"— select —",
"membrane-bound",
"intracellular",
"N/A (not applicable)",
]
# -----------------------------
# Scenarios (titles simplified; type NOT shown in title)
# -----------------------------
SCENARIOS = {
"Cardiomyocyte signaling": {
"secreting": "Cardiomyocyte",
"molecule": "Intracellular ions",
"receiving": "Neighboring cardiomyocytes",
"type": "direct",
"mol_class": "N/A (not applicable)",
"receptor": "N/A (not applicable)",
"explain": {
"type": "Gap junctions are contact dependent → direct signaling.",
"mol_class": "Electrical/ionic coupling across connexons; not a classic ligand.",
"receptor": "No classic receptor: current spreads via channels/pores.",
},
},
"Cancer proliferation": {
"secreting": "Tumor cell",
"molecule": "EGF",
"receiving": "Same cell",
"type": "autocrine",
"mol_class": "hydrophilic",
"receptor": "membrane-bound",
"explain": {
"type": "Cell releases a signal that acts on itself → autocrine.",
"mol_class": "EGF is a protein → hydrophilic → can’t cross the bilayer.",
"receptor": "EGF binds EGFR (RTK) at the membrane.",
},
},
"Synapse signaling": {
"secreting": "Presynaptic neuron",
"molecule": "Neurotransmitter",
"receiving": "Postsynaptic cell",
"type": "paracrine",
"mol_class": "hydrophilic",
"receptor": "membrane-bound",
"explain": {
"type": "Very short-distance diffusion across synaptic cleft → paracrine.",
"mol_class": "Classical neurotransmitters act extracellularly.",
"receptor": "Postsynaptic receptors are membrane proteins (ionotropic/GPCR).",
},
},
"HPA axis": {
"secreting": "Hypothalamus/Pituitary/Adrenal cortex",
"molecule": "Cortisol",
"receiving": "Liver & skeletal muscle",
"type": "endocrine",
"mol_class": "lipophilic",
"receptor": "intracellular",
"explain": {
"type": "Hormone travels via blood to distant targets → endocrine.",
"mol_class": "Cortisol is steroidal → lipophilic → crosses the membrane.",
"receptor": "Steroids bind cytosolic/nuclear receptors → gene transcription.",
},
},
}
# -----------------------------
# Session state helpers
# -----------------------------
def shuffled(opts):
head, rest = opts[0], opts[1:]
random.shuffle(rest)
return [head] + rest
def ensure_state():
if "secret_opts" not in st.session_state:
st.session_state.secret_opts = shuffled(SECRETING_CELLS_BASE[:])
st.session_state.molecule_opts = shuffled(MOLECULES_BASE[:])
st.session_state.recv_opts = shuffled(RECEIVING_CELLS_BASE[:])
st.session_state.type_opts = shuffled(SIGNAL_TYPES_BASE[:])
st.session_state.class_opts = shuffled(MOLECULE_CLASS_BASE[:])
st.session_state.recept_opts = shuffled(RECEPTOR_LOCS_BASE[:])
if "selections" not in st.session_state:
st.session_state.selections = {
"secreting": "— select —",
"molecule": "— select —",
"receiving": "— select —",
"type": "— select —",
"mol_class": "— select —",
"receptor": "— select —",
}
def reshuffle_all():
st.session_state.secret_opts = shuffled(SECRETING_CELLS_BASE[:])
st.session_state.molecule_opts = shuffled(MOLECULES_BASE[:])
st.session_state.recv_opts = shuffled(RECEIVING_CELLS_BASE[:])
st.session_state.type_opts = shuffled(SIGNAL_TYPES_BASE[:])
st.session_state.class_opts = shuffled(MOLECULE_CLASS_BASE[:])
st.session_state.recept_opts = shuffled(RECEPTOR_LOCS_BASE[:])
clear_selections()
def clear_selections():
for k in st.session_state.selections.keys():
st.session_state.selections[k] = "— select —"
def evaluate(sel, key, scenario_key):
if sel == "— select —":
return None, "Incomplete — choose an option."
correct = SCENARIOS[scenario_key][key]
if sel == correct:
return True, "✔"
explain_map = SCENARIOS[scenario_key]["explain"]
why = explain_map.get("type" if key not in ("mol_class", "receptor") else key, "")
return False, f"Expected: {correct}. {why}"
def draw_diagram(selections, results, scenario_label):
fig, ax = plt.subplots(figsize=(9, 4.6))
ax.set_xlim(0, 13)
ax.set_ylim(0, 6.3)
ax.axis("off")
def color_box(x, y, text, ok, blank=False):
w, h = 2.8, 1.0
face = "#FFF8E1" if blank else ("#E8F5E9" if ok else "#FDECEA")
edge = "#FFB300" if blank else ("#2E7D32" if ok else "#C62828")
ax.add_patch(plt.Rectangle((x, y), w, h, fc=face, ec=edge, lw=2))
ax.text(x + w/2, y + h/2, text, ha="center", va="center", fontsize=11)
return (x, y, w, h)
def arrow(start_rect, end_rect, label=""):
x, y, w, h = start_rect
x2, y2, w2, h2 = end_rect
ax.annotate("", xy=(x2, y2 + h/2), xytext=(x + w, y + h/2),
arrowprops=dict(arrowstyle="->", lw=2))
if label:
ax.text((x + w + x2)/2, y + h/2 + 0.15, label, ha="center", fontsize=10)
# top row: secreting → molecule → receiving
s_ok = results["secreting"][0] if results["secreting"][0] is not None else True
m_ok = results["molecule"][0] if results["molecule"][0] is not None else True
r_ok = results["receiving"][0] if results["receiving"][0] is not None else True
s_rect = color_box(0.6, 3.9, f"Secreting Cell/Tissue\n{selections['secreting']}", s_ok, selections['secreting']=="— select —")
m_rect = color_box(4.6, 3.9, f"Molecule\n{selections['molecule']}", m_ok, selections['molecule']=="— select —")
r_rect = color_box(8.6, 3.9, f"Receiving Cell/Tissue\n{selections['receiving']}", r_ok, selections['receiving']=="— select —")
# bottom row: type → class → receptor
t_ok = results["type"][0] if results["type"][0] is not None else True
c_ok = results["mol_class"][0] if results["mol_class"][0] is not None else True
rc_ok = results["receptor"][0] if results["receptor"][0] is not None else True
t_rect = color_box(2.6, 1.6, f"Signaling Type\n{selections['type']}", t_ok, selections['type']=="— select —")
c_rect = color_box(6.6, 1.6, f"Molecule Class\n{selections['mol_class']}", c_ok, selections['mol_class']=="— select —")
rc_rect = color_box(10.6,1.6, f"Receptor Location\n{selections['receptor']}", rc_ok, selections['receptor']=="— select —")
arrow(s_rect, m_rect, "signal")
arrow(m_rect, r_rect, "response")
arrow(t_rect, c_rect)
arrow(c_rect, rc_rect)
ax.text(6.5, 5.9, scenario_label, ha="center", fontsize=12, fontweight="bold")
st.pyplot(fig)
# -----------------------------
# App UI
# -----------------------------
ensure_state()
st.title("Cell–Cell Communication Builder")
left, right = st.columns([1.1, 0.9])
with left:
scenario_label = st.selectbox(
"Scenario",
options=list(SCENARIOS.keys()),
index=0,
key="scenario_select"
)
with right:
shuffle_clicked = st.button("🔀 Shuffle options", use_container_width=True)
if shuffle_clicked:
reshuffle_all()
st.markdown("Build a consistent map of **Secreting cell/tissue → Molecule → Receiving cell/tissue** and choose **Signaling type**, **Molecule class**, and **Receptor location**. Then click **Test**.")
col1, col2 = st.columns(2)
with col1:
st.subheader("Actors")
st.session_state.selections["secreting"] = st.selectbox(
"Secreting Cell or Tissue",
options=st.session_state.secret_opts,
index=st.session_state.secret_opts.index(st.session_state.selections["secreting"])
if st.session_state.selections["secreting"] in st.session_state.secret_opts else 0,
key="sec_dd",
)
st.session_state.selections["molecule"] = st.selectbox(
"Molecule",
options=st.session_state.molecule_opts,
index=st.session_state.molecule_opts.index(st.session_state.selections["molecule"])
if st.session_state.selections["molecule"] in st.session_state.molecule_opts else 0,
key="mol_dd",
)
st.session_state.selections["receiving"] = st.selectbox(
"Receiving Cell or Tissue",
options=st.session_state.recv_opts,
index=st.session_state.recv_opts.index(st.session_state.selections["receiving"])
if st.session_state.selections["receiving"] in st.session_state.recv_opts else 0,
key="recv_dd",
)
with col2:
st.subheader("Mechanism")
st.session_state.selections["type"] = st.selectbox(
"Signaling Type",
options=st.session_state.type_opts,
index=st.session_state.type_opts.index(st.session_state.selections["type"])
if st.session_state.selections["type"] in st.session_state.type_opts else 0,
key="type_dd",
)
st.session_state.selections["mol_class"] = st.selectbox(
"Molecule Class",
options=st.session_state.class_opts,
index=st.session_state.class_opts.index(st.session_state.selections["mol_class"])
if st.session_state.selections["mol_class"] in st.session_state.class_opts else 0,
key="class_dd",
)
st.session_state.selections["receptor"] = st.selectbox(
"Receptor Location",
options=st.session_state.recept_opts,
index=st.session_state.recept_opts.index(st.session_state.selections["receptor"])
if st.session_state.selections["receptor"] in st.session_state.recept_opts else 0,
key="recept_dd",
)
action_col1, action_col2 = st.columns([1,1])
with action_col1:
tested = st.button("✅ Test", type="primary", use_container_width=True)
with action_col2:
cleared = st.button("🧹 Clear / Reshuffle", use_container_width=True)
if cleared:
reshuffle_all()
results = {}
if tested:
keys = ["secreting", "molecule", "receiving", "type", "mol_class", "receptor"]
for k in keys:
results[k] = evaluate(st.session_state.selections[k], k, st.session_state.scenario_select)
incomplete = any(v[0] is None for v in results.values())
n_ok = sum(1 for v in results.values() if v[0] is True)
total = len(results)
if incomplete:
st.warning("◻️ **INCOMPLETE** — make all selections to test consistency.")
elif n_ok == total:
st.success(f"✅ **CONSISTENT** ({n_ok}/{total}) — Nice! Logical mapping.")
else:
st.error(f"⚠️ **INCONSISTENT** ({n_ok}/{total}). Try again—no hints provided.")
draw_diagram(st.session_state.selections, results, st.session_state.scenario_select)
else:
tmp = {k:(None,"") for k in ["secreting","molecule","receiving","type","mol_class","receptor"]}
draw_diagram(st.session_state.selections, tmp, st.session_state.scenario_select)
st.caption("Options are shuffled on load and when you clear/reshuffle, to emphasize reasoning over pattern matching.")