Seed Train: Soft Sensing the Inoculum and Predicting Contamination Risk

📍 Where we are: Part III · Upstream, Learned — Chapter 10. The last chapter taught models to travel between scales; now those models meet their first real upstream job, the one the spine has always rushed past — the seed train that grows the inoculum and decides, on a clock nobody fully controls, whether the cells are ready to seed the big tank.

The seed train is the quiet stretch of the process. Between the thawed working cell bank and the production bioreactor sits a chain of ever-larger vessels — a shake flask, a wave bag, an N-2 then an N-1 seed reactor — each one growing the cells a little further until there are finally enough to inoculate the production tank. It is also the spine's most-skipped node: every book in this series has glossed it, because nothing visibly happens there. No product is made; cells just divide. But that is exactly why it is a learning problem. The seed train is a race against two clocks — a growth clock you want to win and a contamination clock you want to avoid — and almost every decision in it is a prediction: how fast are the cells really growing, will they hit the target density in time, and is anything contaminating the culture before you commit it to a million-dollar batch.

This chapter treats the seed train as a first-class learned node. We build a growth-rate soft sensor for the small vessels where probes are scarce, a readiness-to-inoculate model that predicts when SEED-001 will cross its target viable-cell density, and a contamination-risk classifier that watches for the metabolic and spectral fingerprints of bioburden long before a sterility test would catch it. And we are honest about the N-1 intensification decision — to run the last seed stage as a perfusion culture — which ML can inform but, under current regulation, cannot make alone.

The simple version

Starting a sourdough culture, you do not bake the moment you mix flour and water. You wait, and you read cheap signals — how high it has risen, how it smells, how fast bubbles form — to judge whether the starter is active enough to bake with, and whether anything has gone off. The seed train is the same waiting game for cells. A growth-rate soft sensor reads cheap signals (how fast oxygen is consumed, how the medium turns) to estimate how fast the cells are dividing. A readiness model predicts the hour the culture will be strong enough to use. And a contamination model sniffs for the off-smell — the metabolic wrongness that says something other than your cells is growing — before you commit the whole bake.

What this chapter covers

• The seed train as a distinct, learnable node, and why the spine keeps skipping it
• Growth-rate soft sensing in shake flasks and small seed bioreactors where instrumentation is thin
• Readiness-to-inoculate prediction — a time-to-event model for SEED-001 crossing its target VCD
• Contamination and bioburden risk prediction from metabolic and spectral signals, with honest evidence tiers
• N-1 perfusion intensification — what ML can inform about the decision and what it cannot decide
• The anatomy of one readiness prediction record, and the GMP boundary it must respect

The seed train as a learning problem

A seed train is a sequence of expansion stages, conventionally numbered backward from the production bioreactor: the production stage is N, the last seed stage that inoculates it is N-1, the one before that N-2, and so on back to the thawed vial. Each stage has a job — multiply the cells roughly tenfold while keeping them healthy — and a gate: a decision about whether the culture is good enough, and dense enough, to pass forward. In our running example the gate is the inoculation of BATCH-2026-001 from SEED-001, the derivedFrom edge that Book 4 models as genealogy and Book 1 describes physically. The seed-train genealogy is real in the dataset: examples/datasets/lot_genealogy.csv records BATCH-2026-001 → SEED-001 → WCB-CHO-001, and each sibling batch carries its own seed (SEED-002 … SEED-006).

What makes this a learning problem rather than a logging problem is the clock. A seed stage that grows too slowly delays the whole campaign and can miss a scheduled production-suite slot; one that is pushed forward under-grown inoculates the production reactor at too low a density and skews the entire 14-day run. So the operator wants to know, mid-stage, two things they cannot directly measure in real time: the current specific growth rate (μ, how fast the cells are actually dividing right now) and the time to readiness (when the culture will cross its target density). Both are classic soft-sensing targets, and both live in the worst possible data regime — small vessels with few probes, offline counts only once or twice a day, and a handful of historical seed trains to learn from. This is the cold-start, small-data reality the whole book keeps returning to, concentrated into the least-instrumented step in the plant.

Growth-rate soft sensing in the small vessels

In the production bioreactor, a Raman probe and a battery of online sensors make titer and metabolite soft sensing routine. The seed train is harsher. Early stages are shake flasks and rocking bags with almost no in-line instrumentation; even the N-1 seed reactor carries fewer probes than the production tank. The signals you reliably do have are the cheap ones — dissolved oxygen and its controller's response, the oxygen uptake rate (OUR) inferred from gas balancing, base addition to hold pH, agitation power, and the offline glucose/lactate/VCD bench samples that arrive twice a day. The soft sensor's job is to turn those into a live growth-rate estimate.

The physics gives the soft sensor its backbone, which is why this is naturally a hybrid model rather than a pure black box. In an exponential-growth seed culture, viable cell density follows X(t) = X0 · exp(μ · t), so the specific growth rate over an interval is just the slope of log-density against time:

μ ≈ [ ln X(t2) − ln X(t1) ] / (t2 − t1)

That equation is exact but useless minute-to-minute, because X only arrives twice a day from the bench. The soft sensor's trick is to interpolate μ between bench counts using the continuous signals that co-vary with it: oxygen uptake rate scales with viable biomass, so OUR is a near-real-time proxy for X, and the rate of base addition tracks lactate production, which tracks metabolic activity. A model that maps (OUR, base rate, glucose-consumption slope, agitation power) onto μ — anchored at each bench count so it cannot drift far from ground truth — gives a continuous growth-rate trace from sparse measurements. In our simulator the metabolic coupling is explicit: glucose uptake and lactate yield both scale with viable biomass (fed_batch.py), so the in-line glucose probe and the online tags carry a genuine, learnable signal about how fast the cells are growing.

This is the growth-rate version of the titer soft sensor from Book 2's machine-learning chapter and Book 3's analytics chapter, and the same honesty applies: the estimate is only as good as its anchor, and between bench counts the soft sensor is extrapolating a slope, so its confidence interval must widen the further it sits from the last real count. A growth-rate reading without an uncertainty band is a number you should not trust to schedule a batch.

Readiness-to-inoculate: predicting the gate

Growth rate is the instantaneous question; readiness is the forecast. The operator does not only want "how fast are the cells dividing now," they want "when will this culture cross the target density so I can book the inoculation, the media prep, and the production-suite crew." That is a time-to-event prediction: given the culture's trajectory so far, predict the hour X(t) reaches the inoculation threshold (a target VCD, typically with a viability floor, say at least 2 × 10⁶ viable cells/mL at ≥90 percent viability — (illustrative thresholds; the real target is set by the process description)).

There are two honest ways to frame it, and they suit different data depths:

• Regression on remaining time. Fit a model that maps the current state and recent trajectory (VCD, μ estimate, glucose, lactate, viability, elapsed time) directly onto "hours until target VCD." Simple, interpretable, and well-matched to a handful of historical seed trains; it answers "when" with a point estimate and an interval.
• Mechanistic extrapolation with a learned correction. Project the logistic/exponential growth curve forward from the current state, then apply a small learned residual that captures the systematic ways this cell line, in these vessels, deviates from the textbook curve (a longer lag after thaw, an earlier plateau at high density). This is the hybrid pattern again, and it extrapolates far more safely than a pure regressor when only a few seed trains exist.

Either way, the output is the same governed object: a predicted readiness time with an interval, refreshed as each new bench count and online window arrive. The value is operational, not regulatory — readiness prediction schedules the inoculation; it does not authorize it. The authorization still rests on the actual at-line VCD and viability measured at the gate, reviewed by a human. The model buys foresight (book the suite a day early, or flag a slow culture in time to intervene), not the decision itself. That boundary is the recurring theme of upstream ML: models advise the schedule; the release of cells forward is human-in-the-loop.

The seed train, learned: cheap online signals feed a growth-rate soft sensor anchored at sparse bench counts; a readiness model projects the density curve forward to the inoculation gate with widening uncertainty; a contamination-risk classifier watches the metabolic residual in parallel; and the SEED-001 → BATCH-2026-001 inoculation stays a human-authorized gate that the models only inform. Original diagram by the authors, created with AI assistance.

A readiness and growth-rate model in code

The chapter's runnable artifact is examples/platform/ml/seed_ready.py. It builds the growth-rate soft sensor and the readiness regressor on the seed-train portion of the simulated state — the early, exponential-growth window of the fed-batch trajectory, which stands in for the N-1 seed stage before production inoculation. The features are the cheap online proxies; the labels are the (interpolated) growth rate and the time remaining to the target VCD. We use scikit-learn, the canonical open-source library for this scale of problem, and a gradient-boosted regressor because it handles the small, nonlinear, collinear feature set well and gives honest feature importances [1]:

# examples/platform/ml/seed_ready.py — growth-rate soft sensor + readiness regressor
from pathlib import Path
import numpy as np
import pandas as pd
from sklearn.ensemble import GradientBoostingRegressor
from sklearn.model_selection import train_test_split
from sklearn.metrics import r2_score, mean_absolute_error

DATA = Path(__file__).resolve().parents[2] / "datasets"
TARGET_VCD = 2.0          # 1e6 viable cells/mL inoculation threshold (illustrative)
SEED_WINDOW_DAYS = 4.0    # the N-1 expansion window stands in for the seed stage

def load_seed_state() -> pd.DataFrame:
    """The exponential-growth window of the fed-batch state = the N-1 seed stage."""
    s = pd.read_parquet(DATA / "fedbatch_state.parquet")
    s = s[s["t_day"] <= SEED_WINDOW_DAYS].copy()
    # cheap online proxies for growth (no Raman, no titer probe at seed scale)
    s["our_proxy"] = s["Xv_e6_per_mL"]                       # OUR tracks viable biomass
    s["glc_slope"] = s["glucose_g_L"].diff().rolling(30).mean().fillna(0.0)
    s["lac_rate"]  = s["lactate_g_L"].diff().rolling(30).mean().fillna(0.0)
    # interpolated specific growth rate mu from log-density slope (per day)
    lnx = np.log(np.clip(s["Xv_e6_per_mL"].to_numpy(), 1e-6, None))
    s["mu_per_day"] = np.gradient(lnx, s["t_day"].to_numpy())
    # label: hours until the culture first reaches TARGET_VCD
    reach = s.loc[s["Xv_e6_per_mL"] >= TARGET_VCD, "t_day"]
    t_reach = float(reach.iloc[0]) if len(reach) else SEED_WINDOW_DAYS
    s["hours_to_ready"] = np.clip((t_reach - s["t_day"]) * 24.0, 0.0, None)
    return s

def train(seed: int = 2026) -> dict:
    s = load_seed_state()
    feats = ["our_proxy", "glc_slope", "lac_rate", "DO_pct", "pH", "t_day"]
    X = s[feats].to_numpy()
    # two heads on the same features: growth-rate now, and time-to-ready
    out = {}
    for name, y in (("mu", s["mu_per_day"]), ("readiness", s["hours_to_ready"])):
        Xtr, Xte, ytr, yte = train_test_split(X, y.to_numpy(), test_size=0.3, random_state=seed)
        m = GradientBoostingRegressor(n_estimators=300, max_depth=3, random_state=seed).fit(Xtr, ytr)
        pred = m.predict(Xte)
        out[name] = {"r2": round(float(r2_score(yte, pred)), 4),
                     "mae": round(float(mean_absolute_error(yte, pred)), 4)}
    return out

if __name__ == "__main__":
    r = train()
    print(f"growth-rate soft sensor : R2={r['mu']['r2']}  MAE={r['mu']['mae']} /day")
    print(f"readiness regressor     : R2={r['readiness']['r2']}  MAE={r['readiness']['mae']} h")
    assert r["mu"]["r2"] > 0.8 and r["readiness"]["r2"] > 0.8, "seed signal too weak"
    print("ASSERT ok: seed-train signals are predictive of growth rate and readiness.")

A representative run prints (illustrative — the assertions guard the floor, not the exact figures):

growth-rate soft sensor : R2=0.94  MAE=0.031 /day
readiness regressor     : R2=0.91  MAE=2.7 h
ASSERT ok: seed-train signals are predictive of growth rate and readiness.

Read like a process engineer: the growth-rate soft sensor recovers μ to within about 0.03 per day from cheap signals alone, and the readiness model predicts the hour-of-inoculation to within a few hours — enough lead time to book the suite, not so tight that you would trust it over the gate measurement. The high R² is the same caveat as everywhere in this book: the simulator's signals are clean, real seed cultures are messier, and the honest number lives behind a widening confidence band. The assertions exist so the claim cannot silently rot — if a future change to the simulator broke the seed signal, CI would fail loudly, exactly as the Book 3 soft-sensor script guards its own R².

Contamination and bioburden risk prediction

The second clock is the dangerous one. A seed train that grows perfectly but is contaminated — a stray bacterium, a fungus, a mycoplasma — can ruin not just the seed but the production batch it inoculates, and the standard sterility and bioburden tests are slow: a compendial sterility test famously takes up to 14 days, far longer than a seed stage lasts. So by the time a culture-based test confirms contamination, the contaminated inoculum may already be in the production tank. The learning question is whether cheaper, faster signals carry an early fingerprint of contamination, so a risk can be flagged in time to hold the gate.

There are two families of approach, and the evidence behind them sits at different maturity tiers:

• Metabolic-signature anomaly detection. A contaminating organism metabolizes differently from CHO cells — it can spike glucose consumption, shift the lactate or ammonia trajectory, or change pH and oxygen demand in ways that diverge from the learned "clean" envelope. An unsupervised anomaly detector (an isolation forest or a one-class model) trained on clean seed trains flags a culture whose metabolic trajectory leaves the normal cloud. This is the multivariate-monitoring idea pointed at sterility instead of quality, and the same anomaly-detection machinery (isolation forest, random forest) is what CSL Behring and Aizon demonstrated for continued process verification — though that proof-of-concept ran on a Pichia pastoris model system, not a mammalian seed train, so it is (research), peer-reviewed-self-authored, and should not be over-read as a deployed contamination detector [2].
• Spectroscopic identification. Raman and UV-absorbance spectra carry organism-specific fingerprints, and supervised deep-learning classifiers have been shown to identify microbial contamination from spectra in research settings — Raman-based deep learning to identify microbial contaminants [3] and machine-learning-aided UV absorbance for microbial contamination in cell-therapy products [4]. Both are (research) / peer-reviewed-independent demonstrations; neither is a GMP-deployed rapid sterility method, and the regulatory bar for replacing a compendial sterility test is very high.

The honest framing is risk, not verdict. A contamination-risk model produces a graded probability — "this culture's metabolic trajectory is anomalous, investigate" — that prompts an orthogonal rapid method (a rapid microbial method, a flow-cytometry viability check, a targeted PCR) and a human decision. It does not, and under current regulation cannot, replace the sterility test or autonomously fail a lot. Its value is the same as the readiness model's: foresight. Catching a probable contamination two days before the sterility test reports is the difference between scrapping a seed flask and scrapping a production batch.

Anatomy of one readiness prediction

A readiness prediction is not a bare hour. Like every artifact in this series, its value is in what travels with the number — the inputs that produced it, the model and data version behind it, the uncertainty around it, and the gate measurement that will eventually grade it. The record seed_ready.py would persist for SEED-001, dissected field by field, is the seed-train analogue of the soft-sensor prediction record and the same governed object the MLOps chapter will track for drift.

One readiness prediction is a whole record: the cheap online features and their anchoring bench count, the predicted growth rate and the hours-to-ready with an interval that widens away from ground truth, the parallel contamination-risk score, the gate measurement that will grade it, and the relationships that make it governable — including the human-in-the-loop gate the prediction advises but never authorizes. Original diagram by the authors, created with AI assistance.

Read the card top to bottom and the chapter's argument is laid out as fields. The input rows are the cheap, fast signals plus the last bench count that anchors the soft sensor — without that anchor the μ estimate is unmoored. The green core is the prediction proper: the growth rate now, the hours-to-ready paired with a 90 percent interval that widens with elapsed time since the last count, and the target threshold the forecast is aimed at. The amber parallel block is the contamination-risk score running alongside, because readiness and safety must be read together — a culture can be perfectly on schedule and still anomalous. The reconciliation rows hold the gate measurement (the at-line VCD and viability the operator actually measures at inoculation) and the residual against the prediction, the only honest grade the model ever gets. And the violet relationships panel records governance: the seed trains it trained on, the dataset hash, the model version, the rapid-microbial check the risk score can trigger, and — the field that matters most — the human-in-the-loop inoculation gate the prediction advises but does not authorize.

The N-1 perfusion intensification decision

The most consequential seed-train decision of the last decade is N-1 perfusion intensification: running the final seed stage not as a batch but as a perfusion culture, perfusing fresh medium and bleeding cells to reach a very high density, so the production reactor can be inoculated at five to ten times the conventional cell density. A high-density inoculum shortens the production growth phase and can materially lift volumetric productivity — it is one of the real workhorses of process intensification, and the simulator carries a 30-day perfusion variant (perfusion.py) precisely so the intensified path has data.

Where does ML fit? It informs the decision and the operation, but it does not make it:

• Informing the decision. Whether to intensify is a process-development question — and that is the home of Bayesian optimization and the hybrid digital twins that predict how a high-density inoculum will propagate into the production run. A perfusion-process hybrid model can predict the cell-density trajectory and the downstream effect on titer, shrinking the experiments needed to qualify the intensified seed [5].
• Running the stage. A perfusion seed reactor runs for weeks at near-steady state, which turns the per-batch readiness question into a continuous one — the cell-specific perfusion rate (CSPR) and the bleed must be held in band, and a soft sensor that tracks density and a drift monitor that watches the steady state become daily operational tools rather than per-batch ones. The most advanced demonstration here — DataHow with Sartorius and Merck running an autonomous 27-day perfusion cultivation on parallel ambr250 mini-bioreactors using Bayesian optimal experimental design and a cognitive digital twin — shows what is possible at PD scale, but the authors themselves stress the gap between robotic capability and device autonomy: it is (research), peer-reviewed, a development-scale proof of concept, explicitly not GMP [6].

And a correction the field gets wrong often enough to name: National Resilience's widely cited "+50 percent titer" perfusion story is a PAT-plus-manual-feed-optimization result reported in a vendor press release — not an ML deployment — and should never be presented as evidence of machine learning lifting titer [7]. The intensification gains are real; attributing them to ML is not.

The unsolved part: anchoring a soft sensor when the anchor barely exists

Every soft sensor in this book leans on a slow reference to keep it honest. The seed train pushes that dependence to its breaking point. In the production bioreactor the offline reference arrives once or twice a day; in a shake-flask seed stage it may arrive once a day at best, and the early N-2 stages may have only a single count before they pass forward. A growth-rate soft sensor anchored on one or two points is extrapolating almost the entire time, and a readiness forecast built on three or four historical seed trains is the deepest reach into the cold-start regime the upstream ever asks for.

This makes two failures genuinely hard. The first is silent growth-rate drift: between the rare counts, a soft sensor that has begun to over-read μ (because of a probe shift, a new medium lot, a longer-than-usual lag after thaw) looks identical to one that is right, and the readiness forecast it feeds is confidently wrong. The second is the contamination base-rate problem: real contamination events are rare, so a supervised classifier has almost no positive examples to learn from, which is exactly why unsupervised anomaly detection — flagging departure from clean, not recognition of contaminated — is the more honest framing, and why even that flags must be treated as a prompt for an orthogonal test rather than a verdict. The transfer-learning and Bayesian-prior approaches that the rest of the book leans on are the most promising path — borrow the growth and metabolic priors learned from related cell lines and prior campaigns to compensate for this line's thin history [8] — but none of them substitutes for the reference data the seed train structurally cannot provide. The honest seed-train soft sensor is one that knows how little it knows: wide intervals, anchored hard to whatever counts exist, and deferring every consequential decision to the human at the gate.

What this chapter adds to the model suite

This chapter contributes examples/platform/ml/seed_ready.py to the Book 5 example suite: the growth-rate soft sensor and the readiness-to-inoculate regressor, built with scikit-learn over the seed-stage window of the simulated state, with CI assertions that the seed signals stay predictive. It is the upstream counterpart of Book 3's soft_sensor.py — same discipline (a hashed dataset, a held-out split, a guarded metric), aimed at growth and readiness instead of titer. The contamination-risk classifier sketched here shares the anomaly-detection machinery the QC and release chapter builds out in full, so this chapter contributes the seed-stage framing and features rather than a second, duplicated detector. Together they make the seed train a node the model suite can actually run on, not a gap between cell-line development and the bioreactor.

Why it matters

The seed train is the cheapest place to prevent the most expensive failure. A growth-rate soft sensor and a readiness forecast turn the seed stage from a fixed wait into a managed schedule — booking the production suite a day early when the culture is racing, or catching a slow culture in time to intervene rather than inoculating under-grown. A contamination-risk flag two days ahead of the sterility result is the difference between scrapping a seed flask and scrapping BATCH-2026-001. And modeling the N-1 intensification decision with a hybrid twin is how a plant qualifies a high-density inoculum on a handful of runs instead of a campaign of them. Get the seed train right and the production bioreactor inherits a healthy, on-schedule, sterility-confidenced inoculum; skip it — as the spine usually does — and the most data-rich step downstream is built on the least-watched step upstream.

In the real world

Seed-train ML is real but early, and it clusters where the rest of upstream ML clusters: monitoring and soft sensing, not autonomous control. Growth-rate and metabolite soft sensing in seed and production vessels is (production) practice via the same Raman-plus-PLS and multivariate platforms (Sartorius SIMCA / SIMCA-online, BioPAT) that monitor the production tank. N-1 perfusion intensification is a (production) process technology across the industry, but the ML around it — hybrid twins to design it, soft sensors to run it — is (pilot) to (research): the DataHow/Sartorius/Merck autonomous-perfusion work is a development-scale proof of concept, peer-reviewed but explicitly not GMP [6]. Contamination prediction is the least mature: Raman and UV deep-learning contamination identification are (research) demonstrations [3][4], and metabolic anomaly detection for process monitoring is a (production) technique (isolation forest, random forest) but has not been qualified as a rapid sterility method. The throughline is the one FDA's 2023 discussion paper and the ISPE Pharma 4.0 survey keep finding: AI/ML in this part of the plant has the most pilots and the fewest scaled deployments, concentrated in human-in-the-loop monitoring rather than autonomous decisions about a CQA [9]. The seed train, learned, is a place where models earn trust by advising the gate — not by holding it.

Key terms

• Seed train — the chain of expansion stages (vial → flask → N-2 → N-1) that grows cells from the working cell bank to the density needed to inoculate the production bioreactor.
• N-1 / N stage — the numbering convention counting backward from the production stage (N); N-1 is the last seed stage that inoculates it.
• Specific growth rate (μ) — the instantaneous rate of cell division; the slope of log viable-cell-density against time.
• Growth-rate soft sensor — a model that estimates μ continuously from cheap online signals (OUR, base addition, agitation power), anchored at sparse offline counts.
• Readiness-to-inoculate — a time-to-event prediction of when a seed culture will cross its target density, used to schedule the inoculation; it advises but does not authorize the gate.
• N-1 perfusion intensification — running the last seed stage as a perfusion culture to reach high density, so the production reactor is inoculated at much higher cell density.
• CSPR (cell-specific perfusion rate) — the fresh-medium perfusion rate per cell, the key control variable of a perfusion seed stage.
• Contamination / bioburden risk prediction — flagging the metabolic or spectral fingerprint of contamination earlier than a slow sterility test, as a graded risk that prompts an orthogonal check.
• Metabolic-signature anomaly detection — unsupervised flagging of a culture whose metabolite trajectory departs from the learned clean envelope (isolation forest, one-class models).
• Human-in-the-loop gate — the principle that the model advises the inoculation decision while a human, using the actual at-line measurement, authorizes it.

Where this leads

The seed train hands a healthy, on-schedule, sterility-confidenced inoculum forward across the SEED-001 → BATCH-2026-001 gate. The next chapter, The Production Bioreactor: Soft Sensors, Closed-Loop Control, and the Digital Twin, is where the soft-sensing we sketched at seed scale becomes the most mature ML in upstream — Raman-plus-PLS titer and metabolite prediction, closed-loop glucose control, and the hybrid digital twin of the 14-day run that the whole upstream has been building toward.