AP-1 cargo read-back

Take 32 archived designs to 128 clonal wells, mark each well so sequencing says which well it came from, and call a pass or a fail per well.

Designs: 32 of 72 on the sheet
Route: index PCR
Wells: 128 over 1 picked plate
Selection: kanamycin

Have in hand

AP-1 cargo archive plate one clonal stock a design, 72 of them, sealed and frozen
Barcoded index primer plate the lab's index primer plate IDX-01, prepared once and called for by a run · 1 µM each

Leaves you with

clonal picked plate one well a picked colony, each well one clone
well calls a pass or a fail a well, and which design each well holds

Keep in mind

A design is read back as a name and a count of the pieces it was built from, never as bases, so any clonal stock this lab holds can be read back here.

The bench is sized from the 32 designs read and not from the sheet, so a design the floor leaves out costs no well, no plate and no reagent.

Which designs are read back

This run reads every design built from 2 fragments or more: 32 of 72. The rest stay polyclonal and are never read one design at a time.

Every design is spotted from AP-1 cargo archive plate, which this run reads a copy of and never changes. A design whose spot grows nothing has no clone to read, and is re-transformed before anyone re-synthesises it.

How a well is told from the rest

The index PCR route marks each well: amplify each well with one barcoded primer pair. The other route marks the same wells a different way, and the picking, the pass rule and the reformat are the same either way.

A well's marks are worked out from where the well is rather than looked up, so a demultiplexer can check an address instead of trusting a file. One axis of the address is the plate: this run pools 2 plates, and the marks tell 96 plates apart on one flow cell.

A well is called above 20 reads. The other route's floor is not a stricter or a looser setting of the same scale, so neither number carries over to the other route.

Materials

Name	Supplier	Catalogue	Storage	Per run	Note	Source
25 cm BioAssay plate	Corning	431111		one spot a design	kanamycin; about 2,500 colonies a plate	^[1]
384-well culture plate	Beckman Coulter	c74290		1, one a 384 wells	60 µL low-salt LB with kanamycin a well	^[1]
96-well half-skirted PCR plate	USA Scientific	1402-9700		2, one a quarter of a picked plate		^[2]
Barcoded index primer plate			-20 °C	2 µL a well at 1 µM each, 96 forward and 96 reverse	prepared once as lab stock, by its own protocol; a run calls for it. The SI's plate preparation ends at 0.1 µM, where its own protocol and the article both stamp 1 µM, which is what this takes	^[2]
ThermoPol Reaction Buffer	NEB	B9004	-20 °C	1 µL a reaction, 10X		^[2]
dNTP mix	NEB	N0447	-20 °C	0.2 µL a reaction, 10 mM each		^[2]
Taq DNA Polymerase	NEB	M0267	-20 °C	0.05 µL a reaction	5 U/µL, so one reaction takes 0.25 units	^[3]
DMSO, molecular biology grade	MP Biomedicals	194819		0.4 µL a reaction, 4% (v/v) in the well		^[2]

Equipment: QPix XE Microbial Colony Picker, multichannel pipette on 96-well spacing, Thermocycler taking a 96-well plate, Incubator at 37 °C, A sequencer, and a demultiplexer that can check an address

Caution: Keep the polymerase on ice.

Plates

picked 1 384 wells · 2 kinds · Beckman Coulter #c74290 · one picked colony each in 60 µL low-salt LB with kanamycin · 128 of 384 wells picked, a quarter at a time

quarter 1
quarter 2

index 1 96 wells · 96 kinds · USA Scientific #1402-9700 · one barcoded PCR each, the pair its own address names · plate mark 1; 96 of 96 wells used

index 2 96 wells · 32 kinds · USA Scientific #1402-9700 · one barcoded PCR each, the pair its own address names · plate mark 2; 32 of 96 wells used

Array and pick

1Array 32 designs and grow

Spot each design from its archive plate as its own spot on a 25 cm BioAssay plate.
Grow overnight at 37 °C on kanamycin.

Expected result

One spot a design, at about 2,500 colonies a plate, which is the density picking wants.

Troubleshooting

A spot grows nothing: That design has no clone to read. Re-transform it from the archive before anyone re-synthesises it.

Notes

This run reads back every design in 2 fragments or more: 32 designs. The rest stay polyclonal and are never read one design at a time.
The archive is untouched: this reads a copy of it.

2Pick 4 colonies of each design

Pick 4 colonies a design into 60 µL low-salt LB with kanamycin, with the QPix XE Microbial Colony Picker.
Fill one quarter of each 384-well plate before starting the next.
Grow overnight at 37 °C.

Expected result

128 wells over 1 plate: 128 picked.
Every colony of one design sits on one plate.

Troubleshooting

A design in many fragments gives no clean colony: The chances above say which designs that is likeliest for. Pick more colonies of those designs before the plates are poured, not after.

Notes

4 colonies a design is the only count measured: four gave a clean copy of 343 of 458 genes.^[4]
2 fragments: 16 designs, 100.0% of picks clean
3 fragments: 6 designs, 93.8% of picks clean
4 fragments: 9 designs, 89.2% of picks clean
5 fragments: 1 design, 84.6% of picks clean
A quarter at a time is what makes a part-filled plate give full plates downstream, and what keeps a mark off a plate that is mostly empty.

Mark every well

3Sample the picked plates into index plates

Move 1 µL of each well into its 96-well plate, one quarter of the picked plate a pass.

Sample a quarter of picked 1 into index 1 multichannel pipette · 96 moves · 1 µL each · every other row and column, starting A1 · one well in four, which is the spacing a head built for the index format reads^[2]

96 moves

From	To	µL
picked 1 A1	index 1 A1	1
picked 1 A3	index 1 A2	1
picked 1 A5	index 1 A3	1
picked 1 A7	index 1 A4	1
picked 1 A9	index 1 A5	1
picked 1 A11	index 1 A6	1
picked 1 A13	index 1 A7	1
picked 1 A15	index 1 A8	1
picked 1 A17	index 1 A9	1
picked 1 A19	index 1 A10	1
picked 1 A21	index 1 A11	1
picked 1 A23	index 1 A12	1
picked 1 C1	index 1 B1	1
picked 1 C3	index 1 B2	1
picked 1 C5	index 1 B3	1
picked 1 C7	index 1 B4	1
picked 1 C9	index 1 B5	1
picked 1 C11	index 1 B6	1
picked 1 C13	index 1 B7	1
picked 1 C15	index 1 B8	1
picked 1 C17	index 1 B9	1
picked 1 C19	index 1 B10	1
picked 1 C21	index 1 B11	1
picked 1 C23	index 1 B12	1
picked 1 E1	index 1 C1	1
picked 1 E3	index 1 C2	1
picked 1 E5	index 1 C3	1
picked 1 E7	index 1 C4	1
picked 1 E9	index 1 C5	1
picked 1 E11	index 1 C6	1
picked 1 E13	index 1 C7	1
picked 1 E15	index 1 C8	1
picked 1 E17	index 1 C9	1
picked 1 E19	index 1 C10	1
picked 1 E21	index 1 C11	1
picked 1 E23	index 1 C12	1
picked 1 G1	index 1 D1	1
picked 1 G3	index 1 D2	1
picked 1 G5	index 1 D3	1
picked 1 G7	index 1 D4	1
picked 1 G9	index 1 D5	1
picked 1 G11	index 1 D6	1
picked 1 G13	index 1 D7	1
picked 1 G15	index 1 D8	1
picked 1 G17	index 1 D9	1
picked 1 G19	index 1 D10	1
picked 1 G21	index 1 D11	1
picked 1 G23	index 1 D12	1
picked 1 I1	index 1 E1	1
picked 1 I3	index 1 E2	1
picked 1 I5	index 1 E3	1
picked 1 I7	index 1 E4	1
picked 1 I9	index 1 E5	1
picked 1 I11	index 1 E6	1
picked 1 I13	index 1 E7	1
picked 1 I15	index 1 E8	1
picked 1 I17	index 1 E9	1
picked 1 I19	index 1 E10	1
picked 1 I21	index 1 E11	1
picked 1 I23	index 1 E12	1
picked 1 K1	index 1 F1	1
picked 1 K3	index 1 F2	1
picked 1 K5	index 1 F3	1
picked 1 K7	index 1 F4	1
picked 1 K9	index 1 F5	1
picked 1 K11	index 1 F6	1
picked 1 K13	index 1 F7	1
picked 1 K15	index 1 F8	1
picked 1 K17	index 1 F9	1
picked 1 K19	index 1 F10	1
picked 1 K21	index 1 F11	1
picked 1 K23	index 1 F12	1
picked 1 M1	index 1 G1	1
picked 1 M3	index 1 G2	1
picked 1 M5	index 1 G3	1
picked 1 M7	index 1 G4	1
picked 1 M9	index 1 G5	1
picked 1 M11	index 1 G6	1
picked 1 M13	index 1 G7	1
picked 1 M15	index 1 G8	1
picked 1 M17	index 1 G9	1
picked 1 M19	index 1 G10	1
picked 1 M21	index 1 G11	1
picked 1 M23	index 1 G12	1
picked 1 O1	index 1 H1	1
picked 1 O3	index 1 H2	1
picked 1 O5	index 1 H3	1
picked 1 O7	index 1 H4	1
picked 1 O9	index 1 H5	1
picked 1 O11	index 1 H6	1
picked 1 O13	index 1 H7	1
picked 1 O15	index 1 H8	1
picked 1 O17	index 1 H9	1
picked 1 O19	index 1 H10	1
picked 1 O21	index 1 H11	1
picked 1 O23	index 1 H12	1

Sample a quarter of picked 1 into index 2 multichannel pipette · 32 moves · 1 µL each · every other row and column, starting A2 · one well in four, which is the spacing a head built for the index format reads^[2]

32 moves

From	To	µL
picked 1 A2	index 2 A1	1
picked 1 A4	index 2 A2	1
picked 1 A6	index 2 A3	1
picked 1 A8	index 2 A4	1
picked 1 A10	index 2 A5	1
picked 1 A12	index 2 A6	1
picked 1 A14	index 2 A7	1
picked 1 A16	index 2 A8	1
picked 1 A18	index 2 A9	1
picked 1 A20	index 2 A10	1
picked 1 A22	index 2 A11	1
picked 1 A24	index 2 A12	1
picked 1 C2	index 2 B1	1
picked 1 C4	index 2 B2	1
picked 1 C6	index 2 B3	1
picked 1 C8	index 2 B4	1
picked 1 C10	index 2 B5	1
picked 1 C12	index 2 B6	1
picked 1 C14	index 2 B7	1
picked 1 C16	index 2 B8	1
picked 1 C18	index 2 B9	1
picked 1 C20	index 2 B10	1
picked 1 C22	index 2 B11	1
picked 1 C24	index 2 B12	1
picked 1 E2	index 2 C1	1
picked 1 E4	index 2 C2	1
picked 1 E6	index 2 C3	1
picked 1 E8	index 2 C4	1
picked 1 E10	index 2 C5	1
picked 1 E12	index 2 C6	1
picked 1 E14	index 2 C7	1
picked 1 E16	index 2 C8	1

Expected result

2 index plates, 128 reactions in all.

Notes

One well in four lines up under a head built for the smaller format, so a quarter moves in one pass.

4Amplify each well with its own pair

Caution: Keep the polymerase on ice.

Add 7 µL of the master mix below to each well, which already holds its 1 µL of culture.
Add 2 µL of the pair its address names from the lab's index primer plate IDX-01, for 10 µL a well.
Seal the plate, spin it down, and run the program below.

Index PCR, one well a sample

Reactions mix includes 50% extra

Component	Stock	Final	1 rxn (µL)	Mix for 128 (µL)
ThermoPol Reaction Buffer^[2]	10X	1X	1	192
dNTP mix^[2]	10 mM each	0.2 mM each	0.2	38.4
Taq DNA Polymerase^[2]	5 U/µL	0.25 units	0.05	9.6
DMSO^[2]	100%	4% (v/v)	0.4	76.8
Nuclease-free water^[2]			5.35	1,027
Barcoded primer mix^[2]	1 µM each	0.2 µM each	2	each tube
Overnight culture^[2]			1	each tube
Total			10	1,344 mix only

Put 7 µL of mix in each tube, then add 2 µL Barcoded primer mix, 1 µL Overnight culture.

Index PCR · 1 h 35 min plus ramps^[2]

Step	Temperature	Time	Cycles
Initial denaturation	95 °C	5 min	1
Denature	95 °C	20 s	×10
Anneal	68 → 63.5 °C -0.5 °C a cycle	20 s
Extend	68 °C	2 min
Denature	95 °C	20 s	×25
Anneal and extend	68 °C	2 min	×25
Final extension	68 °C	5 min	1
Hold	4 °C	∞	1

Index PCR

Expected result

One barcoded amplicon a well. A well's own place in the plate names its forward mark — the first well the first mark — and every well of one plate takes that plate's own reverse mark.
One band a well, the design plus about 100 bp: the two marks, and the stretch between the primer sites and the reading frame.

Troubleshooting

A plate reads back far fewer wells than the others: LevSeq traces that to the amplification, not the sequencing: on a run of ten plates three returned under 60% of their variants. Check each pool on a gel before the library prep.

Notes

96 forward marks and 96 reverse reach 9,216 wells, so the pairs already held cover far more than this run needs.
The primer plate is built once as lab stock and a run calls for it; this protocol does not build one.
The first 10 cycles touch down from 68 to 63.5 °C, 0.5 °C a cycle, and 25 more run at 68 °C.
Extension is 2 minutes, what the authors ran for a gene below 1 kb. The published rule is at least one minute a kilobase, so a longer design wants it raised.
The mix is made 1.5 times what the wells take, which is what LevSeq's own per-plate table spells out. Nothing says what the surplus is for.

5Pool and sequence

Pool each index plate on its own and clean the pool up.
Sequence the pool.

Expected result

Reads for 128 wells, every well told from the rest by the marks it carries.

Notes

This read pools 2 plates. One axis of the address is the plate, so the marks tell 96 plates apart on one flow cell.

Call the wells

6Call every well

Demultiplex the reads by address, checking each address rather than trusting a file.
Call a consensus a well, then compare it with that well's design base for base.
Reformat, compacting out the wells that failed.

Expected result

A well read more than 20 times is called; the index PCR route tolerates 10 reads and warns between the two.
A pass matches across the whole designed region: both entry overhangs, the fragment, the stuffer and the barcode.

Troubleshooting

A well's call is the right protein and the wrong barcode: It fails. A barcode that no longer names its member cannot be put right by the linkage read later.

Notes

A well read too thinly carries no verdict and is not a failure, so reformatting does not compact it out.
More than one consensus is mixed, and mixed fails.
A design with no passing well is picked again from the same archive spot before anyone re-synthesises it.

Sources

Qian et al. 2026, Supplementary Information · Nat. Commun. 10.1038/s41467-026-76740-9 · held under reference_docs/ · 2026-10-06 https://doi.org/10.1038/s41467-026-76740-9 cited at Day 2 · Day 3
Long, Y. et al. 2025, LevSeq, Supporting Information · held under reference_docs/ · 2026-10-06 https://doi.org/10.1021/acssynbio.4c00625 cited at step 2 · step 3 · step 1 · step 4 · thermal cycler table
New England Biolabs, Product Specification: Taq DNA Polymerase with ThermoPol Buffer · PS-M0267S/L/X/E v2.0, effective 12 Feb 2020 · plain curl · 2026-10-08 · docs/research/route-b-index-pcr.md https://www.neb.com/en/-/media/catalog/specifications/m/0/m0267s_l_x_e_v2.pdf cited at Concentration
Lund, S. et al. (2024) Highly parallelized construction of DNA from low-cost oligonucleotide mixtures using Data-optimized Assembly Design and Golden Gate. ACS Synth. Biol. 13, 745-751 · docs/research/synthesis-and-assembly.md https://doi.org/10.1021/acssynbio.3c00694 cited at colonies screened per gene