Deciding what earns plate time

Compound library screening software: in silico compound screening that decides what earns plate time

Upload the library, pick a strain panel, and get every compound back ranked by predicted activity with the mechanism named. The plate budget is the scarce thing here, not the compounds.

Library runs from $1,490 a month

SDF or CSV in, ranked list out
10,000 compounds per batch
Structures never used for training

One compound first, then send the file

of 3 left

Compound, by name or SMILES

Strain panel

S. aureus (MRSA) E. coli K. pneumoniae (CRE) A. baumannii P. aeruginosa E. faecium (VRE)

Read-out

Include closest published analogs

No account needed. 3 screens per session.

What comes back

Rows: One per strain in the panel
Per row: MIC band, S I R call, mechanism
Support: Confidence 1 to 4 and analogs
Typical run: About ten seconds

Sample output Ciprofloxacin Fluoroquinolone

Worked example, replaced when you run

Strain	MIC (µg/mL)	Call	Resistance risk
S. aureus (MRSA)	0.5->32	R Resistant	target mutation, grlA and gyrA Fluoroquinolone resistance is widespread in methicillin resistant lineages.
E. coli	<=0.015-0.06	S Susceptible	efflux, AcrAB-TolC in resistant isolates Potent against wild type; qnr carriage and gyrA changes shift the band sharply.
K. pneumoniae (CRE)	0.5->64	R Resistant	gyrA mutation with plasmid qnr Carbapenemase producing isolates almost always carry quinolone resistance too.
A. baumannii	8->64	R Resistant	efflux, AdeABC Constitutive efflux plus target changes leave little room at achievable exposure.
P. aeruginosa	0.25-2	I Intermediate	efflux, MexAB-OprM Borderline: active on many isolates, lost quickly once efflux is derepressed.
E. faecium (VRE)	4->32	R Resistant	target mutation, parC Enterococci are intrinsically poor fluoroquinolone targets.

Why

Ciprofloxacin is a well characterised fluoroquinolone, so the Gram-negative bands are strongly supported by published activity data. The deciding factor across this panel is not target affinity but exposure: efflux in P. aeruginosa and A. baumannii, and acquired target mutation everywhere resistance is already common. Against a modern ESKAPE panel it reads as a Gram-negative agent with two reliable losses.

Closest published analogs

Levofloxacin Broadly similar Gram-negative bands, better Gram-positive coverage
Delafloxacin Retains activity against many ciprofloxacin resistant staphylococci

Strain	MIC (µg/mL)	Call	Resistance risk	Conf.

Why

Closest published analogs

Want the rest of your series? Batch runs are on the paid tiers. screen left in this session. This session is used up.

Computational prediction from published literature. Research use only, not a lab measurement and not clinical guidance.

The short answer

What compound library screening software is for

Compound library screening software predicts activity for every structure in a library and returns the library ranked, so the assay budget goes to the compounds most likely to be worth it. Here the input is an SDF or a CSV of up to 10,000 compounds, the output is one row per compound per strain with a predicted MIC band, an S, I or R call, the resistance mechanism expected to limit it and a confidence level, and the whole file comes back sorted by that prediction.

The reason to do this at all is arithmetic rather than fashion. Compound is cheap next to plate capacity, analyst hours and the strains themselves, and an antibacterial deck bought off the shelf runs to tens of thousands of structures while a project's realistic assay slot count runs to hundreds. Something has to choose. The alternatives are physicochemical filters, which throw away too much, or somebody's intuition, which leaves no written reason behind an exclusion.

At a glance

Input: An SDF or CSV of structures, up to 10,000 in one batch
Output: The library ranked by predicted panel activity, exportable
Decides: Which compounds are worth a plate, and which are not
Grounding: Published MIC records for the closest analogs of each hit

The constraint, in wells

A catalog library does not fit on the plates you have

Broth microdilution needs a doubling dilution series per compound per strain. Take a 6 strain panel and a 10 point series and one compound occupies about 60 wells, so a 384-well plate carries roughly 6 compounds once growth and sterility controls have their share. Run that multiplication against the libraries people actually buy and the problem states itself.

Library size	Wells at 60 per compound	384-well plates	Ranked here for
1,600 compounds	96,000	267	$238.40 on the Program plan
10,000 compounds	600,000	1,667	$1,490.00 on the Program plan
18,000 compounds	1,080,000	3,000	$2,682.00 on the Program plan
30,000 compounds	1,800,000	5,000	$4,470.00 on the Program plan

The plate count is the honest half of the comparison and the dollar column is the cheap half, but notice what the table does not contain: the cost of the compound itself, the analyst time, the media, the strains and the repeat runs that a real campaign always needs. Those are the numbers that make a full deck screen a capital decision, and they are the reason ranking first is worth doing even when the ranking is imperfect. More on how the assay side is built in the high throughput screening reference, including why Gram-negative hit rates stay stubbornly low.

What is on the shelf

The antibacterial screening libraries you can buy today

Four vendors sell a ready-made antibacterial set, and they are not the same product. Two sell diversity picked for the target class, two sell already-active compounds for repurposing and combination work. Sizes and selection methods below come from each vendor's own catalog page.

Library	Compounds	How it was selected	Worth knowing
Enamine Antibacterial Library	30,000	Knowledge-based selection: privileged cores, motifs and natural product-like scaffolds, refined on molecular parameters.	A dedicated Gram-negative sub-library of 6,400 compounds across 80 plates, plus hit resupply and analogs from stock.
Life Chemicals Antibacterial Libraries	18,000+	Two non-overlapping sets, built with 2D fingerprint similarity (Tanimoto above 0.80) and with docking against bacterial targets.	The focused set is about 15,500 compounds across 70 plus bacterial protein targets, the targeted set about 2,700 picked against pneumonia targets.
MedChemExpress Anti-bacterial Compound Library	1,994	Bioactive compounds with already validated antibacterial activity, rather than a diversity set.	Supplied solid or pre-dissolved, mostly at 10 mM, in 96 or 384-well format. Aimed at repurposing and combination work.
TargetMol Anti-bacterial Compound Library	1,600	Antibiotics and antibacterial bioactive compounds with unique structures, NMR and HPLC or LCMS validated.	Powder or pre-dissolved DMSO in 96 or 384-well plates, optional 2D barcodes. Research use only.

None of the four publishes a price, so every one of them is a quote cycle before you can compare anything. The size column is the part worth staring at: a 30,000 compound deck is roughly 5,000 plates of full panel MIC work, which is why almost nobody screens one end to end and why the deck usually gets cut by something before it gets cut by an assay.

How a library run goes

Four steps, file in to ranked list out

01. Send the file

An SDF or a CSV of SMILES, up to 10,000 structures in one batch. Compound names come along if you have them, and the batch is checked against your remaining month before anything is queued.

02. Pick the panel

ESKAPE, Gram-negative, Gram-positive, or a custom panel of your own strains from Discovery upward. The panel choice is the single biggest lever on what the ranking means, so pick the organisms the program actually has to beat.

03. Read the ranking

Every compound comes back with a predicted MIC band per strain, an S, I or R call, the mechanism expected to limit it and a confidence level. Sort by predicted activity, or filter to compounds where no efflux mechanism is flagged.

04. Export and plate the top

CSV, SDF or a PDF report, so the ranked list goes into a registry, a cherry-pick order or a project meeting. What you plate is now a decision with a written reason behind every exclusion.

Teams that run this continuously rather than in sittings usually move it into their own code instead, which is what the antibacterial screening API is for: the same batch endpoint, called by a nightly job rather than by a person with a file.

Who runs libraries through this

Four teams with the same bottleneck

A biotech that just bought a deck

The library arrived, the plate budget did not grow to match. Ranking the deck against the target panel turns 18,000 compounds into a first cherry-pick that can be defended to a board.

A medicinal chemistry team with enumerated analogs

Enumeration produces thousands of plausible analogs a week. Scoring them against the panel before anything reaches a synthesis queue is the cheapest decision in the whole cascade.

A discovery CRO quoting a campaign

Clients ask what a screening campaign will find. A ranked library and a mechanism breakdown make the proposal concrete, and the same export becomes the deliverable.

An academic AMR lab on a fixed grant

The grant bought a finite number of plates. Spending them on a prediction-enriched subset rather than a uniform sample of the deck is the difference between a paper and a shrug.

What it will not do

The limits, before you send a deck

A ranking is only useful if you know what it is ranking on. None of this is hidden until after a card is charged.

It ranks, it does not measure

Every row is a prediction grounded in published MIC records for the closest characterised analogs. Broth microdilution against your isolates stays the ground truth. The ranking decides what earns that plate.

Novel scaffolds come back low confidence on purpose

A structure with no characterised relatives has no honest band, and the row says so rather than averaging into something confident. In a library run that is a routing signal: those compounds are candidates for a plate precisely because prediction cannot help.

A full deck is more than one month

Program covers 10,000 screens a month with batches up to 10,000 compounds. A 30,000 compound deck is three passes or a larger allowance, and the batch is refused up front rather than half run.

Potency is not a development candidate

Predicted activity against a strain says nothing about permeability in your hands, exposure, clearance or toxicity. Those are separate problems and are not claimed here.

Questions people ask before buying

Library screening, answered plainly

How many compounds are typically screened in the discovery phase?

For antibacterials it ranges from a focused set of about 2,000 bioactive compounds up to whole-deck campaigns of hundreds of thousands. The commercial antibacterial libraries sit in between: 1,600 to 30,000 compounds. The number that actually matters is not the library size, it is how many of them your assay slots can hold, which is usually one or two orders of magnitude smaller.

How do you screen a compound library for antibacterial activity?

Physically, by broth microdilution: each compound goes into a doubling dilution series against each strain, and the lowest concentration that stops visible growth is the MIC. Computationally, the library is scored first and only the top of the ranked list is plated. Most teams now do both in that order, because plate capacity is the constraint and the ranking is far cheaper than the plate.

What does compound library screening software actually do?

It takes a file of structures, predicts activity for every one of them against the strains you care about, and hands back the library sorted by that prediction with a reason attached to each row. On this site that means a predicted MIC band per organism, an S, I or R call, the resistance mechanism expected to limit the compound, and a confidence level so thin rows are visible as thin.

Is virtual screening better than high throughput screening?

They answer different questions and work best in sequence. High throughput screening measures, which nothing computational replaces, but it costs plates, compound and time. Virtual screening ranks, cheaply and imperfectly, which makes it the right way to decide what goes on the plate. Using prediction as the filter and the assay as the confirmation is what keeps a small budget productive.

How large a library can I screen in one batch?

Up to 10,000 compounds in a single batch on the Program plan, 1,000 on Discovery and 50 on Lab. A 10,000 compound batch also uses the whole Program month, so a 30,000 compound deck is either three monthly passes or a conversation about a larger allowance. The batch is refused up front if the remaining month cannot cover it, rather than half run.

What does it cost to screen a compound library?

Between $0.15 and $0.60 per compound against a full six-strain panel, depending on the plan, because the price is published and the allowance is fixed. Ranking a 10,000 compound library on the Program plan is $1,490 for the month. Compare that against what the same 10,000 compounds would cost in plates, compound handling and analyst time.

Do I still need to run the assay?

Yes, and the software is worth less if you pretend otherwise. The prediction decides which compounds earn a plate. Broth microdilution against your own isolates stays the ground truth, and the value of ranking first is that the plates you do run are enriched for actives instead of being a uniform sample of a deck.

Can I screen my own proprietary compounds, not a catalog library?

That is the normal case. Structures you upload stay yours, are not shared and are not used to train models, which is the practical reason an unpublished series can go through a hosted tool at all. Enumerated analogs from a design step are handled the same way: send the file, get the ranked list back, decide what to make.

Pricing

What ranking a library costs per compound

Full plan comparison

Published, because a quote cycle costs a small team weeks it does not have. Program is $1,490 a month for 10,000 screens, about $0.15 per compound against a full panel, with batches up to 10,000. Discovery is $499 for 2,000 screens and batches up to 1,000, which suits a series rather than a deck. The single compound case is the same engine through the MIC prediction software, and peptide libraries go through the antimicrobial peptide prediction tool with sequences as the input. The full breakdown, including how the other pricing models in this market bill, is in the screening software pricing write-up, and how the wider drug discovery software market discloses its prices, vendor by vendor, is on the drug discovery software cost page.

Yearly billing is two months cheaper, charged once a year.

Lab

$124 /mo

$1,488 charged once a year

250 screens a month against the standard panels, with CSV export.

Get started

Academic AMR lab, 1 to 2 chemists

Discovery

Try one compound before you send the file

Run a single structure against the panel here, then upload the library on a plan that covers it.

Run a screen first

What you are agreeing to

Your compounds stay yours
Never used for model training
Deletable on request
No card needed to create your account