Estrus and Data Science: Implications to the Pre-implantation Endometrium Biology and Genomic Selection

我的主要研究目标是最大限度地利用精密传感器技术的大数据和构建工具，揭示泌乳奶牛子宫内膜与发情强度相关的特定生物学机制，并提出提高奶牛发情表达、胚胎存活率和遗传选择的策略。在最近的一系列研究中，使用不同的自动活动监测系统，与发情强度低的奶牛相比，发情强度高的奶牛的生育能力提高了35%。这一新颖的观点是，只有在最近几年，在奶牛场采用传感器技术的情况下，才能在大量奶牛上详细准确地显示发情的相对增长、持续时间和总体模式。使用峰值强度、持续时间和其他数字测量可以帮助预测生育率，并改善乳制品行业使用的生殖计划的决策。此外，使用自动化监测系统作为一种客观的数字工具来选择具有优越生育力的动物，同时提高表型收集的精度，确实有潜力。数据科学工具和人工智能现在有能力探索来自活动监视器的大型数据集，以便找到与生育能力和其他关键生理事件相关的更精确的测量值。迄今为止，使用动情表达特征作为基因组选择表型的文献有限。选择与发情数字特征相关的遗传标记可以加速选择生育能力优越的奶牛，减少乳制品行业对生殖激素的依赖，并提高数据科学能力，在精密乳制品中创建数字表型。为了回答一些与动情表达相关的机制问题，我们选择研究着床前阶段的子宫内膜，这一阶段以胚胎丢失率高而闻名。本课题组前期研究发现，受子宫内膜发情检测影响的转录本属于免疫系统和粘附分子家族（如MX1、MX2、MYL12A、MMP19、CXCL10、IGLL1、SLPI、PTX3、IDO、MUC1、MUC4、SELL），以及与前列腺素合成相关的转录本（ERa、OTR、COX-2）。研究表明，发情的奶牛在丝状阶段的受胎大小增加了2倍。我们的小组对这三种生物功能特别感兴趣，它们由大约150个转录本组成，因为它们在发情期间不断被修改，应该使用更多样化和更复杂的细胞和分子生物学技术进一步深入研究。然而，迄今为止还没有研究使用与发情行为强度对子宫内膜机制功能相关的数字表型的详细信息。

The major aim of my research program is to maximize the use of big data from precision sensor technologies and construct tools to unveil specific biological mechanisms in the endometrium of lactating dairy cows associated with estrus strength, and to propose strategies that enhance estrous expression, embryonic survival, and the genetic selection of dairy cows. In a series of recent studies using different automated activity monitoring systems, cows displaying estrus events of greater intensity had a 35% increase in fertility compared with estrus events of low intensity. The novel proposition is that detailed and accurate relative increase, duration and overall pattern of estrus display have only been possible on large number of cows in recent years with the adoption of sensor technologies in dairy farms. The use of peak intensity, duration, and other digital measurements could assist in the prediction of fertility and improve decision-making of reproductive programs used in the dairy industry. Moreover, there is true potential to use automated monitor systems as an objective digital tool to select animals of superior fertility, while improving phenotype collection precision. Data science tools and artificial intelligence have now the capability to explore large datasets coming from activity monitors in order to find more precise measurements associated with fertility and other key physiological events. To date, there is limited literature using estrous expression characteristics as phenotypes for genomic selection. The selection of genetic markers associated with digital characteristics of estrus could accelerate the selection of cows with superior fertility, decrease reliance on reproductive hormones in the dairy industry, and improve the data science capabilities to create digital phenotypes in precision dairy. In order to answer some of the mechanistic questions associated with estrous expression, we chose to study the endometrium during the pre-implantation stage, a period known for the high rates of embryonic loss. In previous studies by our group we found that transcripts affected by estrous detection in the endometrium belong to the immune system and adhesion molecule family (e.g. MX1, MX2, MYL12A, MMP19, CXCL10, IGLL1, SLPI, PTX3, IDO, MUC1, MUC4, SELL), as well as those related with prostaglandin synthesis (ERa, OTR and COX-2). The studies have shown a 2-fold increase in conceptus size at the filamentous stage coming from cows that displayed estrus. Our group has been particularly interested in those three biological functions, comprised of around 150 transcripts, as they have been consistently modified by estrus, and should be further investigated in depth using a more varied and sophisticated cellular and molecular biology techniques. No study to date, however, have been performed using detailed information of digital phenotypes related with the behavioural strength of estrus on mechanistic functions of the endometrium.