呵呵,这是别人问的,我得到他的允许可以把原信及回复贴在这里.关于徐老师的文章及新书介绍这篇博文里有http://blog.sciencenet.cn/home.php?mod=space&uid=479743&do=blog&id=391882

原邮件:  

徐老师：
    您好！
    非常感谢您能在百忙之中抽出时间阅读我的邮件。
    前段时间拜读了您编写的《molecular plant breeding》，深受启发，也坚定了我
将来从事育种研究的决心。
    我即将成为...研究所硕博二年级学生。研究生一年级的时候，我阅读了大量关联
分析相关的文章，对这个方向非常感兴趣，也准备做这方面的研究。但是当前遇到了一
个很大的问题：不知道该如何构建关联群体。文章上只是笼统的说选择具有代表性的材
料。可是不知道怎样评估一些材料是否有代表性，更不知道怎样才能建立一个有代表性
的群体。所以非常希望能得到您的指点。另外，我想做...的关联分析，但是听说国内
也有很多科研单位在做，而且他们的材料都很丰富，这个问题也困惑了我很久，感觉万
一跟他们做重复了就白做了，您有没有好的建议啊
    再次衷心的表示感谢~
祝您天天好心情

答复:

Have been very busy recently.

Here are some brief points for you to consider:

1. For small-effect QTL, association mapping will have no advantage at all.
We will find a lot of QTL but we can do nothing with them. Just for
publication. If you go for them, you do not need to worry as you can always
publish in a journal by the end.(回复中的黑体字是我自己加的)
2. For large-effect QTL, it is hard to have a complete story for publication
in a high IF journal. Association mapping has to be combined with other
approaches including biparental population-based linkage mapping and others.
Joint linkage-LD mapping is a choice. We have a recent paper in PNAS about
this, which addressed the strategies: parallel and integrated joint mapping.(这篇文章本博客有:关联作图和连锁作图的结合http://blog.sciencenet.cn/home.php?mod=space&uid=479743&do=blog&id=378810)
3. Does not matter how good an AM panel is. It can be only good for some
traits, if you want to clone the genes by the end.
4. The key is the population size (better with over 500 accessions) and they
comes from different origins and breeding programs etc.
5. For genes with rare alleles, AM will not work. So be sure your panel is
segregating normally for the traits of interest.
6. The more important the trait of interest is, the higher chance to have
strong competition. So if you work on a hot trait, you have to be ready for
all kinds of competition, e.g. four groups in the world work on the same
genes at about the same rate.

All the recent review papers will not tell you many truth about the
disadvantages of AM. You will have to learn by practice or by doing.
Sometimes you will know them too late.

So many people are interested in AM as an independent approach. We will
repeat soon the history of mapping QTL with small effects: finding thousands
of QTL but doing nothing for real breeding.

Hope this is helpful.

Yunbi