PLEASANTON — November 13, 2026 -- The Molecular Biology Core Facilities (MBCF) at Dana-Farber Cancer Institute have added n6 Tec's icon96 system to their targeted cell-free DNA (cfDNA) sequencing service. MBCF reports that the coefficient of variation (CV) in library yield across 10 ng samples fell from 54% with fixed-cycle amplification to 8% with n6's AutoNorm software.
That is a drop of 46 percentage points, or about 85% by our calculation. The figures come from an n6 application note and a SelectScience webinar in which MBCF reported its results. They are not independently published data.
Yield variability falls from 54% to 8% when each reaction stops on its own endpoint
The icon96 gives each of its 96 wells independent thermal control and real-time amplification monitoring. With AutoNorm, each reaction stops when it reaches a sample-specific endpoint. Amplification, quantification and normalization happen in the same run, according to n6.
Fixed-cycle programs push every sample through the same number of cycles. According to the release, low-input samples can then under-amplify below the yield needed for capture, while higher-input samples over-amplify. The over-amplified samples produce excess duplicates and need extra quantification, dilution and normalization before pooling.
At 1 ng, the CV in library yield was 30%, about 3.8 times the 8% measured at 10 ng. The 1 ng and 10 ng samples ran together. After icon96 joined the workflow, on-target rates exceeded 90% across four sample sets evaluated at 1 ng and 10 ng, MBCF reported.
Workflow detects 0.5% variants from 1 ng, roughly 1.5 mutant molecules
The release puts 1 ng of cfDNA at approximately 300 genome equivalents. A 0.5% allele frequency in that input works out to about 1.5 variant-carrying copies per target, by our calculation.
In tests with Twist reference standards, MBCF reported that all spiked variants were reliably detected down to 0.5% allele frequency at both 1 ng and 10 ng. Mean target coverage at 1 ng was above 200x. The release notes that rare tumor-derived variants often sit below 1% allele frequency, so the 0.5% result is the figure that matters for that use.
Patient-derived samples spanning 2 ng to 25 ng run together in a single batch
MBCF normalized patient-derived cfDNA samples of approximately 2 ng to 25 ng, a 12.5-fold input range, within one run. All of them generated enough library to proceed to hybrid capture.
For investigators who submit samples to the service, that means mixed-input batches no longer need manual rebalancing, according to the release. MBCF Director Zach Herbert said investigators bring "precious samples where every molecule counts." The release does not say how many samples or runs underlie the CV figures.
The service combines the IDT xGen cfDNA & FFPE DNA Library Preparation Kit, dual-UMI duplex sequencing, amplification on icon96, hybrid capture and sequencing on an Illumina NovaSeq X Plus.
n6 CEO and co-founder Pranav Patel said the workflow adapts to each sample rather than forcing a whole batch through one process, and that this simplifies operations. n6 products are for research use only and not for use in diagnostic procedures.