🧬 biology

Evaluate the performance of Targeted Sequencing for Tumor FFPE Samples and ctDNA Standards on the Sikun 2000

This study demonstrates that the Sikun 2000 sequencing platform delivers superior quality metrics, high reproducibility, and enhanced sensitivity for detecting low-frequency variants in tumor FFPE samples and ctDNA standards compared to established Illumina platforms, establishing it as a competitive alternative for clinical and research applications.

Original authors: Jiuzhou Zhao, Tianbo Wang, Li Zhao, Jiawen Zheng, Chengzhi Zhao, Junnan Feng, Rui Sun, Chengjiao Duan, Dongdong Chen, Keya Cai, Bing Wei

Published 2026-08-06

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Jiuzhou Zhao, Tianbo Wang, Li Zhao, Jiawen Zheng, Chengzhi Zhao, Junnan Feng, Rui Sun, Chengjiao Duan, Dongdong Chen, Keya Cai, Bing Wei

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

Imagine the human body as a massive, bustling city. Sometimes, a few buildings in that city start to malfunction and turn into trouble spots—these are tumors. To stop them, doctors need a map of exactly what's gone wrong inside the cells. For a long time, the only way to get this map was to send a team in with a shovel to dig up a piece of the broken building (a tissue biopsy). But sometimes, the building is too hard to reach, or the damage is too small to see with a shovel.

Enter "liquid biopsy." Instead of digging, doctors can look for tiny, floating scraps of the building's blueprint that the trouble spots have accidentally dropped into the city's bloodstream. These scraps are called ctDNA. The problem is, these scraps are incredibly rare and fragile, like finding a single specific grain of sand on a beach. To find them, scientists use a high-tech machine called a Next-Generation Sequencer (NGS). Think of this machine as a super-fast, ultra-precise photocopier that reads the genetic code. The goal is to make a machine that is not only fast and accurate but also affordable, so more people can get these life-saving maps.

This paper is a report card for a new photocopier called the Sikun 2000, made by a company called Sikun Life Science. The researchers wanted to see if this new machine could do the job as well as, or even better than, the famous "gold standard" machines currently in use (like the Illumina NextSeq 550 and NovaSeq series). They tested the Sikun 2000 on two things: real tumor samples taken from patients (FFPE samples) and "practice" samples with known, tiny amounts of mutations (ctDNA standards).

Here is what they found:

The Real-World Test (Tumor Samples)
The team took 50 real tumor samples and ran them through both the new Sikun 2000 and the established NextSeq 550. They wanted to see if the new machine could read the genetic code clearly and find the same mistakes as the old one.

The Result: The Sikun 2000 did an excellent job. It produced clearer, sharper "photos" of the genetic code, with a higher quality score (called Q30) than the NextSeq 550. It also managed to read the unique parts of the code more deeply (higher unique depth), which is crucial for spotting rare errors.
The Match: When they compared the lists of mutations found by both machines, they agreed on almost everything (about 96% of the time). The measurements of how common a mutation was (VAF) were also nearly identical. This suggests the Sikun 2000 is a reliable replacement for the older machines when looking at real patient tissue.

The "Needle in a Haystack" Test (ctDNA Standards)
Next, they tested the machine's ability to find very rare mutations. They used special "training wheels" samples (standards) that contained known mutations at very low levels: 0.1% and 1.0%. Finding a 0.1% mutation is like finding one specific red grain of sand in a bucket of a thousand white ones.

The Result: The Sikun 2000 was a champion at this. It found 100% of the mutations in the 0.1% and 1.0% samples.
The Comparison: When they compared the Sikun 2000 to two other high-end Illumina machines (NovaSeq 6000 and NovaSeq X), the Sikun 2000 actually performed slightly better. It found all 8 expected mutations in the difficult 0.1% sample, while the other two machines missed one of them (the KRAS A146T mutation). The Sikun 2000 also produced fewer "blurry" or low-quality reads, meaning its data was cleaner.

What This Means
The paper suggests that the Sikun 2000 is a strong contender in the world of cancer genetics. It offers high-quality data that matches or beats the current top-tier machines, especially when trying to spot very rare mutations in liquid biopsies. The authors note that while the machine is great for targeted sequencing (looking at specific genes), it might not be the best choice for massive, whole-genome projects due to its output limits. However, for the specific job of finding cancer mutations in patient samples, the Sikun 2000 shows significant potential to be a competitive, high-quality, and cost-effective tool for both research and clinical use.

Technical Summary: Evaluation of Targeted Sequencing for Tumor FFPE Samples and ctDNA Standards on the Sikun 2000

Problem Statement
Next-generation sequencing (NGS) has become the standard for detecting tumor gene variants to guide precision oncology. However, as sequencing scales expand, there is an increasing demand for platforms that offer both high data quality and lower costs. While Illumina platforms (e.g., NextSeq 550, NovaSeq 6000) remain the mainstream standard, domestic alternatives are emerging. The Sikun 2000, a new sequencing platform based on sequencing by synthesis (SBS) with a reversible terminator approach, has been launched for scientific research. It claims compatibility with Illumina libraries, high accuracy, good repeatability, and lower operational costs. However, its performance relative to established Illumina platforms in clinical tumor applications—specifically regarding FFPE tissue samples and low-frequency circulating tumor DNA (ctDNA)—required rigorous evaluation.

Methodology
The study employed a comparative experimental design to evaluate the Sikun 2000 against established Illumina platforms (NextSeq 550, NovaSeq 6000, and NovaSeq X).

Sample Preparation:
- FFPE Samples: 50 tumor FFPE samples from cancer patients were processed using a standardized DNA extraction and library construction protocol (Burning Rock Lung Plasma® Multi-Gene Mutation Detection Kit with a 168-gene panel).
- ctDNA Standards: Lung cancer ctDNA standards with known Variant Allele Frequencies (VAF) of 0.0% (negative control), 0.1%, and 1.0% were prepared using the IGT® Fast Library Prep Kit and TargetSeq® probes (NSCLS 23 gene panel).
Sequencing:
- FFPE samples underwent triplicate sequencing on the Sikun 2000 and single sequencing on the NextSeq 550.
- ctDNA standards were sequenced on the Sikun 2000, NovaSeq 6000, and NovaSeq X.
- All platforms utilized 150 bp paired-end (PE150) sequencing.
Data Processing and Analysis:
- Raw data was downsampled to equivalent depths for fair comparison.
- Quality control included filtering for adapter contamination, low-quality bases, and N-content.
- Reads were aligned to the hg19 reference genome using BWA-MEM.
- Variant calling (SNVs, Indels) was performed using VarScan, followed by annotation with ANNOVAR and SnpEff.
- Statistical analyses (Wilcoxon signed-rank test, One-Way ANOVA, Pearson correlation) were used to compare quality metrics (Q30, unique depth, mapping ratio) and variant detection concordance.

Key Contributions and Results

Performance on FFPE Samples (Sikun 2000 vs. NextSeq 550):
- Quality Metrics: The Sikun 2000 demonstrated superior performance in unique sequencing depth and Q30 ratios compared to the NextSeq 550. The Q30 ratio for Sikun 2000 ranged from 95.6% to 97.5%, significantly higher than the NextSeq 550's 88.8%–95.9% ( $P < 0.01$ ). The unique depth was also significantly higher for Sikun 2000 ( $P < 0.01$ ).
- Reproducibility: Triplicate sequencing on the Sikun 2000 showed high reproducibility, with key quality parameter curves exhibiting a high degree of overlap.
- Variant Concordance: Both platforms showed high consistency in detecting driver gene variants (EGFR, ALK, KRAS, etc.) in NSCLC, with a concordance rate of 95.83%. The Variant Allele Frequencies (VAF) of SNVs and Indels showed a strong correlation ( $R^2 \geq 0.96$ ) between the two platforms.
Performance on ctDNA Standards (Sikun 2000 vs. NovaSeq 6000/X):
- Sensitivity: For ctDNA standards with 0.1% and 1.0% VAF, the Sikun 2000 achieved 100% detection sensitivity for all 8 mutation sites.
- Comparative Sensitivity: At the 0.1% VAF level, the Sikun 2000 outperformed the NovaSeq 6000 and NovaSeq X. While Sikun 2000 detected all 8 sites, the NovaSeq platforms detected only 87.5% (failing to detect the KRAS p.A146T SNV).
- Data Quality: The Sikun 2000 generated significantly fewer low-quality reads (0.05%) compared to NovaSeq 6000 (0.65%) and NovaSeq X (0.86%) ( $P < 0.05$ ). It also achieved a higher average target depth (28,355×) compared to the NovaSeq platforms.

Significance and Claims
The paper claims that the Sikun 2000 represents a competitive alternative for tumor-targeted sequencing in both scientific research and clinical applications. The study concludes that the Sikun 2000 offers:

Superior Data Quality: Higher Q30 ratios and unique depths compared to the NextSeq 550 in FFPE samples.
Enhanced Low-Frequency Detection: Superior sensitivity for low-frequency variants (0.1% VAF) compared to high-throughput Illumina NovaSeq platforms, attributed to lower rates of low-quality reads and higher target depth.
High Reproducibility: Consistent results across repeated assays.
Cost-Effectiveness: The authors note the platform's significantly lower running costs compared to Illumina platforms, suggesting strong market competitiveness.

The authors position the Sikun 2000 as a viable option for precision medicine, particularly for detecting driver gene variants in NSCLC and monitoring ctDNA, while acknowledging that broader clinical adoption will require further validation with larger clinical sample datasets.

More like this