mosbot

mosbot is a web app for mosquito activity experiments: plot LD/DD actograms and combine experiments from activity CSVs.

Built for circadian pipeline.

Live app: https://mosbot.streamlit.app

---

Features

Section	What it does
Activity graphs	Actograms, LD/DD profiles, day/night totals, and stats from an activity CSV
Combine experiments	Merge several experiments onto one ZT timeline — kinds line up, each keeps its true ZT start + LD/DD schedule, gaps become NaN — then plot combined Figs 3–9 + stats

---

Quick start

Requirements: Python 3.10–3.12

git clone https://github.com/florawwang/mosbot.git
cd mosbot
pip install -r requirements.txt
./run_app.sh

Open http://127.0.0.1:8502 and enter the lab passcode.

Passcode (keep this out of git)

The unlock code is not stored in the public repo. Configure it one of these ways:

1. Local file (recommended for laptop):

   cp .streamlit/secrets.toml.example .streamlit/secrets.toml
   # edit secrets.toml and set CLOUD_VIEWER_PASSCODE

2. Environment variable:

   CLOUD_VIEWER_PASSCODE=yourcode ./run_app.sh

3. Streamlit Cloud: App → Settings → Secrets → paste:

   CLOUD_VIEWER_PASSCODE = "yourcode"

Ask a lab member for the current passcode.

Stop the app with Ctrl+C. If port 8502 is busy: PORT=8503 ./run_app.sh

For YOLO inference and detection-cache builds (heavier deps):

pip install -r requirements-ml.txt

---

What files you need

Input	Description
Raw images	Folder of timelapse frames (.jpg / .png)
Labels CSV	MakeSense well boxes (one row per mosquito)
YOLO model	Detector weights (e.g. uninf_det_v0.pt) — local inference only

After inference you also get:

• activity_transposed.csv — per-frame movement per mosquito

Set paths in the app sidebar (or use env vars / auto-discovery when nested in the lab data repo).

---

Typical workflow

1. Run inference (laptop or VM — can take hours on CPU) → activity CSV
2. Open mosbot → plot activity graphs and combine experiments

Skip step 1 if someone already shared an activity CSV with you.

Inference (local)

From this repo:

export PYTHONPATH="."
pip install -r requirements-ml.txt

python -m mosquito_lab.run_inference \
  --image-folder "/path/to/raw_images" \
  --labels "/path/to/labels.csv" \
  --model "/path/to/uninf_det_v0.pt" \
  --output-dir ./mosquito-lab-output \
  --serve-viewer

Background run:

./run_cloud.sh local "/path/to/raw_images" "/path/to/labels.csv" "/path/to/model.pt"

Deploy (Streamlit Community Cloud)

Hosted at https://mosbot.streamlit.app.

Setting	Value
Main file	streamlit_app.py
Python	3.10–3.12

Inference is meant for a laptop/VM. On Cloud, upload saved activity CSVs.

If the app fails to start, open Manage app → Logs on share.streamlit.io.

---

Troubleshooting

Problem	Fix
App won't start	pip install -r requirements.txt; use Python 3.10–3.12
Port already in use	PORT=8503 ./run_app.sh
Empty graphs	Load an activity CSV in the sidebar, or run inference first
Wrong passcode	Ask a lab member, or set CLOUD_VIEWER_PASSCODE