wangpeng407 / reporterscore Goto Github PK

After downloading the software, you will use the data contained in the folder, which will also report an error
通过git clone下载完，用文件夹里面自带的KO.rela.xls和group.list运行，也会报错

my command:
perl -MList::Util=sum -lane'$.==1 && print ; $.==1 && next; sum(@f[1..$#F]) == 0 && next; print ' KO.rela.xls > non-zero-KO.rela.xls
Rscript caculate_reporter_score.R --group group.list --abund non-zero-KO.rela.xls --type pathway --outdir try

here is the error :
报错长这样：
$ Rscript caculate_reporter_score.R --group group.list --abund non-zero-KO.rela.xls --type pathway --outdir try
Error in combn(unique_group, 2) : n < m
Calls: as.data.frame -> combn
Execution halted

尊敬的作者您好，我是一个小白
我在用您的log.sh构建库的过程中，发现有很多No KOs in map，可以请您更新一下log.sh或者是告诉我是怎么回事吗？
非常感谢！

perl handle.pl pathway-KO.list pathway.desc.list > path_stat_KO.xls
No KOs in map01010
No KOs in map01060
No KOs in map01061
No KOs in map01062
No KOs in map01063
No KOs in map01064
No KOs in map01065
No KOs in map01066
No KOs in map01070
No KOs in map03260
No KOs in map03263
No KOs in map03264

目前我是直接写的github的网站

Hi, I noticed that in function StatKOAttributes, using the adjusted p-value from Wilcoxon rank-sum test or t-test to calculated Z-score of every KO. I wonder what if I used p-value generated from other statistical method? For example, microbiome data typically is overdispersed and sparse with many zeros, so we use metagenomeSeq package using zero-flated  Gaussian distribution to perform differential analysis instead of Wilcoxon rank-sum test or t-test. In that case, is it suitable that use the p-value calulted form metagenomeSeq package to generated z-sore?

最近有一篇文章讨论了reporter-score正负号误用问题：https://mp.weixin.qq.com/s?__biz=MzUzMjA4Njc1MA==&mid=2247507105&idx=1&sn=d5a0f0aaf176e245de7976f0a48f87a8#rd

其中就提到了你所参考的1. Bäckhed, F. et al. Dynamics and Stabilization of the Human Gut Microbiome during the First Year of Life. Cell Host & Microbe 17, 690–703 (2015). 也存在误把reporter-score正负号当作上下调标志的现象。

但是我仔细阅读了你的代码，发现你巧妙的将矫正后p值除以2从而让z-score全部小于0，再通过判断上下调给予z-score正负号，从而让最后的reporter-score正负号作为上下调标志
然而我看backhed 2015的method里没有写到这样的做法，想问一下你是参考哪里的方法这样做呢？

另外想问一些问题：
1.如果输入的ko丰度全部上调或者下调，这会让z-score全部小于0或者大于0，但是最后的reporter-score还是会有正负号，因为random_mean_sd_from_vec得到的大概是整体ko的均值和方差，这样的reporter-score正负号就不能代表上下调，跟最近那一篇文章提到的一样。
2.random_mean_sd_from_vec这个函数，如果clean.KO长度小于KOnum的话，会使得每次采样的结果一样（不放回取样），得到的sd理论上是0，实际是很小的浮点数，最终reporter-score会非常大，有考虑这种情况使用放回抽样吗？

Hi,I notice that column "type " in "stat.ko.res.xls" show the "KO“ enrich or deplete in CG. However the "reporter_score.xls" don't show the type of the "pathway". I wonder how can I confirm the type of the pathway ?

Hi,I wonder that "path_stat_KO" and "module_stat_KO" list are an universal data for the ReporterScore analysis? Or I should construct my own "path_stat_KO" and "module_stat_KO" data acccording to my own data. But how?