请教QC-RLSC和QC-RFSC

[asepsiswu]

关于预测值，statTarget2/R/shiftCor_v2.R代码的是rfP <- predict(temp, y)和yf <- stats::predict(loe, y)。

我理解predict的数值(rfp或yf)就是校正后的数值

后面x[i, ] <- as.numeric(x[i, ])/rfP 和 x[i, ] <- as.numeric(x[i, ])/yf 以及 loessDatT <- loessDatmp * 1000 代码的含义是什么。

比较了shift_QC_cor.csv与shift_QC_raw.csv的QC同一样本的feature值，差别极大，如何理解 测试代码如下：

datpath <- system.file('extdata',package = 'statTarget') samPeno <- paste(datpath,'MTBLS79_sampleList.csv', sep='/') samFile <- paste(datpath,'MTBLS79.csv', sep='/') shiftCor(samPeno,samFile, Frule = 0.8, MLmethod = 'QCRLSC', QCspan = 0,imputeM = 'KNN',plot=TRUE)

[13479776]

Hi asepsisWu,

You can refer my recently published paper

“statTarget: A streamlined tool for signal drift correction and interpretations of quantitative mass spectrometry-based omics data. Analytica Chimica Acta. dio: https://doi.org/10.1016/j.aca.2018.08.002 https://doi.org/10.1016/j.aca.2018.08.002"

Simply, We use the predicted value as the adjusted factor.

The second question, The difference between raw value and corrected value does not matter. Because we use the relative Quan. but not the absolute Quan. approach.

Thanks for your attention to our statTarget.

Best regards, Hemi Luan, Ph.D. Research assistant professor SUSTech Academy for Advanced Interdisciplinary Studies Southern University of Science and Technology 1088 Xueyuan Rd., Shenzhen, China E-mail: luanhm@sustc.edu.cn / hm-luan@msn.com

在 2019年3月13日，上午11:04，asepsisWu notifications@github.com 写道：

关于预测值，statTarget2/R/shiftCor_v2.R代码的是rfP <- predict(temp, y) https://github.com/statTarget/statTarget2/blob/56f161c86ae1544f358828c1da463fbc89b1aaf7/R/shiftCor_v2.R#L260和yf <- stats::predict(loe, y) https://github.com/statTarget/statTarget2/blob/56f161c86ae1544f358828c1da463fbc89b1aaf7/R/shiftCor_v2.R#L297。

我理解predict的数值(rfp或yf)就是校正后的数值

后面x[i, ] <- as.numeric(x[i, ])/rfP https://github.com/statTarget/statTarget2/blob/56f161c86ae1544f358828c1da463fbc89b1aaf7/R/shiftCor_v2.R#L261 和 x[i, ] <- as.numeric(x[i, ])/yf https://github.com/statTarget/statTarget2/blob/56f161c86ae1544f358828c1da463fbc89b1aaf7/R/shiftCor_v2.R#L298以及 loessDatT <- loessDatmp * 1000 https://github.com/statTarget/statTarget2/blob/56f161c86ae1544f358828c1da463fbc89b1aaf7/R/shiftCor_v2.R#L362 代码的含义是什么。

比较了shift_QC_cor.csv与shift_QC_raw.csv的QC同一样本的feature值，差别极大，如何理解 测试代码如下：

datpath <- system.file('extdata',package = 'statTarget') samPeno <- paste(datpath,'MTBLS79_sampleList.csv', sep='/') samFile <- paste(datpath,'MTBLS79.csv', sep='/') shiftCor(samPeno,samFile, Frule = 0.8, MLmethod = 'QCRLSC', QCspan = 0,imputeM = 'KNN',plot=TRUE)

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/statTarget/statTarget2/issues/4, or mute the thread https://github.com/notifications/unsubscribe-auth/ASTTMF0KjUKuXCBD4aH5ENt2QdWbYXT-ks5vWGrPgaJpZM4bsS_Z.

[asepsiswu]

我认为，yf作为预测值，sample.norm也作为预测值，sample.norm/yf的比值 == sample值/QC值，sample.norm 的值应该是乘以yf。 从线性回归的角度考虑，假设QC的拟合曲线是 y = 1.2 X，那么 sample的预测值是sample.raw 1.2，即sample.norm = QC.norm / QC.raw sample.raw 这种情况， x[i, ] <- as.numeric(x[i, ])/QC.median[i]yf 能否更加详细的解释x[i, ] <- as.numeric(x[i, ])/yf 背后的含义？

[asepsiswu]

statTarget的论文 提到x'ij =xij / Fij。一般情况下都是x'ji = xij * 系数。

[13479776]

It is a good new idea. I have not chance to try it. I am looking for the results. If you have any progress, please do not hesitate to notice me. Thanks.

Best regards, Hemi Luan, Ph.D. Research assistant professor SUSTech Academy for Advanced Interdisciplinary Studies Southern University of Science and Technology 1088 Xueyuan Rd., Shenzhen, China E-mail: luanhm@sustc.edu.cn / hm-luan@msn.com

在 2019年3月13日，下午2:49，asepsisWu notifications@github.com 写道：

statTarget的论文 提到x'ij =xij / Fij。一般情况下都是x'ji = xij * 系数。

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/statTarget/statTarget2/issues/4#issuecomment-472301346, or mute the thread https://github.com/notifications/unsubscribe-auth/ASTTMNF_kJjh_AYDPVXKpJiZ8BfegzXcks5vWJ-OgaJpZM4bsS_Z.