Laboratory of Genetics and Molecular Cardiology/LIM 13, Heart Institute-InCor, University of São Paulo Medical School, São Paulo, São Paulo, Brazil.
PLoS One. 2013 Aug 8;8(8):e70605. doi: 10.1371/journal.pone.0070605. eCollection 2013.
The a priori identification of induced pluripotent stem cells remains a challenge. Being able to quickly identify the most embryonic stem cell-similar induced pluripotent stem cells when validating results could help to reduce costs and save time. In this context, tools based on non-classic logic can be useful in creating aid-systems based on visual criteria. True colonies when viewed at 100x magnification have been found to have the following 3 characteristics: a high degree of border delineation, a more uniform texture, and the absence of a cracked texture. These visual criteria were used for fuzzy logic modeling. We investigated the possibility of predicting the presence of alkaline phosphatase activity, typical of true induced pluripotent stem cell colonies, after 25 individuals, with varying degrees of experience in working with murine iPS cells, categorized the images of 136 colonies based on visual criteria. Intriguingly, the performance evaluation by area under the ROC curve (16 individuals with satisfactory performance), Spearman correlation (all statistically significant), and Cohen's Kappa agreement analysis (all statistically significant) demonstrates that the discriminatory capacity of different evaluators are similar, even those who have never cultivated cells. Thus, we report on a new system to facilitate visual identification of murine- induced pluripotent stem cell colonies that can be useful for staff training and opens the possibility of exploring visual characteristics of induced pluripotent stem cell colonies with their functional peculiarities. The fuzzy model has been integrated as a web-based tool named "2see-iPS" which is freely accessed at http://genetica.incor.usp.br/2seeips/.
诱导多能干细胞的预先鉴定仍然是一个挑战。在验证结果时，能够快速识别最类似于胚胎干细胞的诱导多能干细胞，可以帮助降低成本和节省时间。在这种情况下，基于非经典逻辑的工具可以用于创建基于视觉标准的辅助系统。在 100 倍放大倍数下观察到的真正集落具有以下 3 个特征：高度的边界描绘、更均匀的质地和不存在裂纹质地。这些视觉标准用于模糊逻辑建模。我们研究了在 25 名具有不同程度的小鼠诱导多能干细胞工作经验的个体中，预测碱性磷酸酶活性（典型的真实诱导多能干细胞集落）存在的可能性，根据视觉标准对 136 个集落的图像进行分类。有趣的是，通过 ROC 曲线下面积的性能评估（16 名个体具有令人满意的性能）、Spearman 相关性（均具有统计学意义）和 Cohen's Kappa 一致性分析（均具有统计学意义）表明，不同评估者的区分能力相似，即使是那些从未培养过细胞的评估者。因此，我们报告了一种新的系统，以促进对小鼠诱导多能干细胞集落的视觉识别，这对于员工培训很有用，并为探索诱导多能干细胞集落的视觉特征及其功能特性提供了可能性。模糊模型已被整合为一个名为“2see-iPS”的基于网络的工具，可在 http://genetica.incor.usp.br/2seeips/ 免费访问。
