How much of Thermo Fisher’s antibody data has been manipulated?

赛默飞世尔（Thermo Fisher）有多少抗体数据被篡改了？

Research Integrity 研究诚信

How much of Thermo Fisher’s antibody data has been manipulated? 赛默飞世尔（Thermo Fisher）有多少抗体数据被篡改了？

Published by Reese Richardson on May 28, 2026 由 Reese Richardson 发布于 2026 年 5 月 28 日

[ TL;DR: As of 3 June 2026, we have identified more than 450 images bearing signs of manipulation in verification data advertised by Thermo Fisher Scientific in its online primary antibodies catalog (+1 by Abcam). See the full repository of problematic images, curated by myself and Sholto David, here: Zenodo – Problematic images in vendor antibody verification data. You are welcome to contribute new findings at this Google form. This blog post was original posted on 28 May 2026 and has not been edited to update counts since that date. There is an update covering Thermo Fisher’s response at the bottom of this post. ] [ 长话短说：截至 2026 年 6 月 3 日，我们在赛默飞世尔科技（Thermo Fisher Scientific）在线一抗目录中宣传的验证数据里，发现了超过 450 张带有篡改迹象的图片（另有 1 张来自 Abcam）。由我和 Sholto David 整理的问题图片完整存储库请见此处：Zenodo – 供应商抗体验证数据中的问题图片。欢迎通过此 Google 表单贡献新的发现。本博文最初发布于 2026 年 5 月 28 日，自该日期起未对统计数字进行更新。本文底部有关于赛默飞世尔回应的更新内容。 ]

A week and a half ago, while looking for trustworthy data demonstrating a cell line’s deficiency in the protein p53, Sholto David came across the following image of a Western blot in Thermo Fisher Scientific’s online antibodies catalog: 一周半前，Sholto David 在寻找证明某细胞系缺乏 p53 蛋白的可信数据时，在赛默飞世尔科技的在线抗体目录中偶然发现了以下这张蛋白质印迹（Western blot）图片：

A Western blot presented as “Advanced Verification” data for an anti-p53 monoclonal antibody. This image is supposed to demonstrate that the antibody being sold works as intended. It is labeled as “Advanced Verification” data on Thermo Fisher’s site and its caption implies that the data was produced internally (other images in the catalog that have not been produced internally are labeled under “Published Figures”). 这是一张作为抗 p53 单克隆抗体“高级验证”（Advanced Verification）数据展示的蛋白质印迹图。该图片旨在证明所售抗体能按预期发挥作用。它在赛默飞世尔网站上被标记为“高级验证”数据，其说明文字暗示该数据为内部产生（目录中非内部产生的数据被标记为“已发表图表”）。

This Western blot appears to be fabricated. As annotated by Sholto, several of the bands in the image are identical after flipping and rotation: The bands labeled 1 through 4 are all identical to one another after a vertical flip, a horizontal flip or a 180 degree rotation. 这张蛋白质印迹图似乎是伪造的。正如 Sholto 所标注的那样，图中的几个条带在经过翻转和旋转后完全相同：标记为 1 到 4 的条带在经过垂直翻转、水平翻转或 180 度旋转后，彼此完全一致。

Shortly after, Johan Duchêne noticed a similarly suspicious image of another anti-p53 antibody in Thermo Fisher’s catalog. I decided to go looking myself and quickly turned up ten more suspicious images on eight other antibody products offered by Thermo Fisher. 不久之后，Johan Duchêne 在赛默飞世尔的目录中发现了另一张同样可疑的抗 p53 抗体图片。我决定亲自查看，并很快在赛默飞世尔提供的其他八种抗体产品中又发现了十张可疑图片。

Sholto and I have now documented more than 100 images provided as verification data in Thermo Fisher’s antibody catalog that have apparently been manipulated. You can see all of them at this Zenodo repository, which we’ll try to update regularly. This repository also contains a handful of instances that are less suggestive of manipulation, but the data is still problematic (e.g., the same image being presented as verification data for two different antibodies). 我和 Sholto 目前已经记录了超过 100 张在赛默飞世尔抗体目录中作为验证数据提供、且明显经过篡改的图片。你可以在这个 Zenodo 存储库中查看所有这些图片，我们将尽量定期更新。该存储库还包含少数不太像是篡改的案例，但数据仍然存在问题（例如，同一张图片被用作两种不同抗体的验证数据）。

Here are some highlights: 以下是一些重点：

Some images are similar to the example that started this excursion and also feature bands that are unusually similar to one another. 有些图片与引发此次调查的例子类似，其条带之间也存在异常的相似性。

Many images, if you adjust the contrast, feature conspicuous “brushstrokes”, suggesting that part of the image has been painted over in a program like Photoshop. 许多图片在调整对比度后，会出现明显的“笔触”，这表明图片的部分内容是在 Photoshop 等程序中被涂抹过的。

Other images feature repetitive blocks of background noise, suggesting that parts of the image were copy-pasted over each other. They might also feature sudden unexpected discontinuities in the pattern of background noise. 其他图片则呈现出重复的背景噪声块，表明图片的部分内容是相互复制粘贴的。它们还可能在背景噪声模式中出现突兀且不自然的断层。

In one instance, I thought I had stumbled across another one of these instances of duplicated blocks of background noise… …only to discover that dozens of antibodies for sale from Thermo Fisher present a verification Western blot that features this exact background pattern, just with minimal edits such that the single band is positioned where one would expect to see the protein of interest. 有一次，我以为自己又偶然发现了这种背景噪声块重复的案例……结果却发现，赛默飞世尔出售的数十种抗体所展示的验证蛋白质印迹图，都具有完全相同的背景模式，只是经过了极小的编辑，使得单一的条带出现在人们预期看到目标蛋白的位置。

A slideshow of contrast-adjusted Western blots all featuring “background pattern A”. 一组经过对比度调整的蛋白质印迹图幻灯片，它们都具有“背景模式 A”。

At the time of writing, we’ve documented 50 instances of this background pattern appearing in verification data on Thermo Fisher’s site, but this is far from an exhaustive list. “Similar image” searches using Google Lens, Bing Images or DuckDuckGo betray hundreds more that we have yet to document. 在撰写本文时，我们已经记录了 50 例这种背景模式出现在赛默飞世尔网站验证数据中的情况，但这远非全部。使用 Google Lens、Bing Images 或 DuckDuckGo 进行“相似图片”搜索，还会发现数百个我们尚未记录的案例。

Antibodies are near-ubiquitous but notoriously fickle laboratory reagents in biomedical research. For many applications, it is absolutely crucial that the antibodies that you use are selective (i.e., the antibody binds strongly to the target protein) and specific (i.e., the antibody binds to the protein of interest and little else). 抗体在生物医学研究中几乎无处不在，但也是出了名的“善变”的实验室试剂。对于许多应用而言，所使用的抗体必须具有选择性（即抗体与目标蛋白强结合）和特异性（即抗体仅与目标蛋白结合，而不与其他物质结合），这一点至关重要。

Commercially-available antibodies often fail to meet these criteria. Members of YCharOS, an independent antibody validation initiative, estimated in 2024 that “more than 50% of all antibodies failed in one or more applications”. Antibodies that don’t work as intended can delay experiments by weeks and non-specific antibodies are a massive source of irreproducibility in the biomedical literature. 市售抗体往往无法满足这些标准。独立抗体验证计划 YCharOS 的成员在 2024 年估计，“超过 50% 的抗体在一种或多种应用中失效”。无法按预期工作的抗体会导致实验延误数周，而非特异性抗体则是生物医学文献中不可重复性的巨大来源。

To learn more, check out Johan’s September 2025 talk in which he details his experience with a study published using a non-specific antibody. 欲了解更多信息，请查看 Johan 在 2025 年 9 月的演讲，他在演讲中详细介绍了自己使用非特异性抗体发表研究的经历。

Antibody vendors like Thermo Fisher (probably the largest laboratory reagent supplier in the world) put verification data in their catalogs to demonstrate to scientists that the product works as intended. While signs of manipulation in this verification data don’t necessarily imply that the antibodies in question don’t work as advertised, without reliable verification data available, scientists will have no way of knowing until they have actually purchased the antibody. 像赛默飞世尔（可能是全球最大的实验室试剂供应商）这样的抗体供应商在目录中提供验证数据，是为了向科学家证明产品能按预期工作。虽然这些验证数据中的篡改迹象并不一定意味着相关抗体无法按广告宣传的那样工作，但在没有可靠验证数据的情况下，科学家在实际购买抗体之前将无从知晓。

And antibodies are not cheap; at Thermo Fisher, a single vial containing a 0.1 mL aliquot of antibody solution typically costs 400 to 500 USD. 而且抗体并不便宜；在赛默飞世尔，单瓶 0.1 毫升的抗体溶液通常售价为 400 到 500 美元。

We created our repository of problematic images in vendor antibody catalogs A) to raise awareness among working biomedical scientists that the antibody verification data they see in a vendor’s catalog may be unreliable and B) to encourage others to look for and report problematic vendor-provided antibody verification data (not limited to just Thermo Fisher). If you spot anything, feel free to fill out this Google form so that it might be added to the spreadsheet and repository. 我们创建供应商抗体目录问题图片存储库的目的在于：A) 提高生物医学科研人员的意识，让他们意识到在供应商目录中看到的抗体验证数据可能是不可靠的；B) 鼓励其他人寻找并报告供应商提供的有问题抗体验证数据（不仅限于赛默飞世尔）。如果你发现了任何问题，请随时填写此 Google 表单，以便将其添加到电子表格和存储库中。

A parting message: always validate your antibodies! 临别寄语：请务必验证你的抗体！

UPDATE 8 June 2026: Thermo Fisher has released a galling 15-point response to our observations. The most important part (in my assessment) is quoted below (emphasis mine): 6. Did Thermo Fisher manipulate or fabricate antibody data?No. 更新 2026 年 6 月 8 日：赛默飞世尔针对我们的观察结果发布了一份令人恼火的 15 点回应。其中最重要的部分（据我评估）引用如下（重点由我标注）：6. 赛默飞世尔是否篡改或伪造了抗体数据？没有。