Bishop D T, Sherman S L
Genet Epidemiol Suppl. 1986;1:123-8. doi: 10.1002/gepi.1370030719.
The complete likelihood analysis is required to obtain the evidence for the gene order. Since the number of candidate orders increases rapidly with the number of loci and the complete analysis requires considerable computer time, a method for rapidly screening order is required. To this end, we used the program MAP83 for the initial investigation of order. This method provided a subset of orders to be further investigated by the complete likelihood method. We used PAP for these subsequent analyses. The advantage of algorithms such as MAP83 is that the basic unit of analysis is the set of pairwise LOD scores which are routinely calculated at the beginning of a linkage analysis. However, the magnitude of the log likelihood differences between orders can only be suggestive for two reasons. First, the usual assumptions for MAP83 regarding independence of the recombination fraction estimate are not met by these data and, therefore, no attempt can be made to interpret the log likelihood differences statistically. Secondly, pairwise information precludes identifying the specific inheritance pattern of each gamete, thus the information content of obligatory multiple recombinants is lost. For example, the order HBBC-ADJ-HRAS1-INS requires five double recombinants for the Utah data while the order HBBC-ADJ-INS- HRAS1 does not require any (White et al, 1985). The support for the former order over the latter is a log likelihood difference of 3.73 for the complete likelihood analysis when the HBBC locus is ignored.(ABSTRACT TRUNCATED AT 250 WORDS)
为了获得基因顺序的证据,需要进行完整的似然性分析。由于候选顺序的数量会随着基因座数量的增加而迅速增加,并且完整的分析需要大量的计算机时间,因此需要一种快速筛选顺序的方法。为此,我们使用程序MAP83进行顺序的初步研究。该方法提供了一组顺序,以便通过完整似然性方法进行进一步研究。我们在后续分析中使用了PAP。诸如MAP83之类算法的优点在于,分析的基本单位是成对LOD分数集,这些分数在连锁分析开始时通常会被计算出来。然而,顺序之间对数似然性差异的大小仅具有提示性,原因有两个。首先,这些数据不满足MAP83关于重组分数估计独立性的通常假设,因此无法尝试对对数似然性差异进行统计学解释。其次,成对信息无法识别每个配子的特定遗传模式,因此强制多次重组的信息内容丢失了。例如,对于犹他州的数据,顺序HBBC-ADJ-HRAS1-INS需要五个双重组体,而顺序HBBC-ADJ-INS-HRAS1则不需要任何双重组体(怀特等人,1985年)。当忽略HBBC基因座时,完整似然性分析中前一个顺序相对于后一个顺序的支持是对数似然性差异为3.73。(摘要截断于250字)