paperKB
coga / coga-kb
Help
Sign in

Chunk #11 — Results — Effects of different filtering thresholds

Source
Effects of filtering by Present call on analysis of microarray experiments.
Embedded
yes

Text

when adjusted to remove a comparable number of probe sets, leaves a less favorable balance between Absent and Present probe sets. Unlike filtering by fraction Present, signal-based filtering removes Present probe sets even when using a very low minimum MAS5 signal: a minimum MAS5 signal of 254 leaves the same number of probe sets as the Present filter set at >0%; at that signal threshold, 2.7% of the Present probe sets are removed along with 66% of the Absent probe sets. For RMA the balance between removing Absent and Present probe sets is even worse: the equivalent threshold of 4.25 removes 12.1% of Present probe sets and 57% of Absent. As the signal value is increased beyond that, a greater fraction of Present probe sets is removed. Average Signal or RMA value thresholds that leave a number of probe sets equal to the results of using 50% Present remove 11.3% and 21.6% of Present probe sets while only removing 86% and 77% of the Absent probe sets (for MAS5 signal or RMA value, respectively; Figs. 3B, 3C).