In the odd case that the person who has the maximum "days to event" value also has an "event" value of 1, the variance for MANTEL's log-rank test is being calculated with a (1-1) denominator and returning -nan values for all test statistics. When this happens the output can't compare test statistics and reports all the genes of having a best threshold at 0.
scwest
commented
5 years ago
In the odd case that the person who has the maximum "days to event" value also has an "event" value of 1, the variance for MANTEL's log-rank test is being calculated with a (1-1) denominator and returning -nan values for all test statistics. When this happens the output can't compare test statistics and reports all the genes of having a best threshold at 0.