Skip to content

Learn by Family

In-depth, step-by-step tutorials. They are organized by graph family, so pick the track for the graph you chose in Which Graph Should I Use?. The two tracks are independent; you can follow either or both.

LZGraph track

The general-purpose family: gene-aware, scalable, ML-ready.

1. Graph Construction

Beginner · 15 min

Build AAP, NDP, and Naive graph variants from your data, with gene annotations and abundance weighting.

2. Sequence Analysis

Beginner · 20 min

Score sequences with pgen, explore graph structure, and simulate new sequences.

3. Diversity Metrics

Intermediate · 15 min

Measure repertoire complexity with k-diversity, Hill numbers, and occupancy models.

FlashBackGraph track

The Markovian family: exact, sampling-free analytics and anomaly scoring.

1. Exact Diversity

Compute Hill numbers and effective diversity exactly via forward dynamic programming, and understand why "exact" matters.

2. Anomaly Detection

Score sequences for surprise with SCALE, the self-calibrated anomaly score, and interpret the result.

3. Personalization & Algebra

Bayesian posterior updates, leave-donor-out construction, and graph algebra (union, intersection, difference).

Prerequisites

Sample data

The LZGraph-track tutorials use example data included with LZGraphs:

import csv

with open("examples/data/ExampleData1.csv") as f:
    sequences = [row['cdr3_rearrangement'] for row in csv.DictReader(f)]

Next steps