From 2131bcf69aeb4bacdd0ac441f041f795be410075 Mon Sep 17 00:00:00 2001 From: PAMatusiak Date: Sun, 28 Nov 2021 20:40:01 -0500 Subject: [PATCH 1/2] Paige's attempted code review --- vignettes/TET2.Rmd | 48 +++++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 45 insertions(+), 3 deletions(-) diff --git a/vignettes/TET2.Rmd b/vignettes/TET2.Rmd index a79f410..cf3e31c 100644 --- a/vignettes/TET2.Rmd +++ b/vignettes/TET2.Rmd @@ -12,6 +12,10 @@ vignette: > --- ```{r setup, include=FALSE} + +BiocManager::install("GEOquery") +BiocManager::install("limma") + knitr::opts_chunk$set(echo = TRUE) knitr::opts_chunk$set(collapse = TRUE, comment = "#>") library(devtools) @@ -45,6 +49,9 @@ details can be found in [the paper](https://doi.org/10.1016/j.ccr.2010.11.015), which is indeed a landmark. Nevertheless, some fine details of the work seemed to disagree with the results of other cohorts when replication was attempted. +#To be accurate, all of the 2021 cohort members were very much so alive prior +#to 2010 + ![The money shot](figure/TET2.png) Some of you who have seen volcano plots before can guess where this is going. @@ -77,10 +84,16 @@ if (FALSE) { # this takes about 5 minutes: # how many probes, how many patients? dim(DNAme) -# Features Samples -# 25626 394 - +# Features Samples +# 25626 394 ``` +#^^note:features==probes, samples==patients +#^^this output indicates that only 25,626 CpG sites are available for analysis, +#though the original paper reports testing ~50,000 sites. Why is there such a +#discrepancy? +#original paper excludes 15 of 398 patients for analysis, yet there are still +#394 patients included in the dataframe. Why are there not 385 patients +#included, as reported in the paper? ### Some contrasts @@ -151,3 +164,32 @@ fit3 <- eBayes(lmFit(exprs(DNAme)[, as.integer(rownames(design3))], design3)) # 10 probes for TET2:purity ``` +#Paige's code review: + ##The following additions were necessary in the first chunk of code: + #1) BiocManager::install("GEOquery") + #2) BiocManager::install("limma") + +#Paige's science review: + 1) Assumptions regarding the term "probe": + A) I assume a "probe" is one of the ~50,000 CpG islands sites + that the authors looked at in the paper. + B) I also assume that "6513 probes for IDH" (for example) infers that + the same 6513 CpG islands are mutated across all samples harboring an + IDH mutation. + 2) According to the heatmap provided (under #always plot your data), + it is evident IF either IDH1/2 or TET2 mutations are present in the AML + patients, these mutations LARGELY occur mutually exclusive of each other. + However, it should be noted that most of the samples harbored neither mutation. + Additionally,the heatmap also demonstrates a small population of samples that + harbor both mutations.Therefore, it is misleading to say these mutations are + mutually exclusive. + A) Question: If the vast majority of the samples extracted from these AML + patients do not harbor either mutation, is this really a defining + feature of AML? + 3) The 6500+ hits for methylated CpG sites in IDH mutated-samples vs. + the ~10 hits for the TET2 mutated-samples indicates that these mutations have + differential associations with hypermethylation at CpG sites. It is misleading + for the authors to indicate that both kinds of mutations engender "overlapping" + hypermethylation signatures, considering that even if all TET2-related sites + overlapped with the IDH-related sites, this only represents 0.15% of all + sites implicated by IDH mutation. \ No newline at end of file From fa5854e4e69b84d6fd662ad8ea874a29cbb8874c Mon Sep 17 00:00:00 2001 From: PAMatusiak Date: Sun, 28 Nov 2021 21:00:02 -0500 Subject: [PATCH 2/2] Paige's attempted code review #2 --- vignettes/TET2.Rmd | 46 ++++++++++++++++++++++++++++++++++------------ 1 file changed, 34 insertions(+), 12 deletions(-) diff --git a/vignettes/TET2.Rmd b/vignettes/TET2.Rmd index cf3e31c..380c9a0 100644 --- a/vignettes/TET2.Rmd +++ b/vignettes/TET2.Rmd @@ -88,12 +88,12 @@ dim(DNAme) # 25626 394 ``` #^^note:features==probes, samples==patients -#^^this output indicates that only 25,626 CpG sites are available for analysis, +#^^This output indicates that only 25,626 CpG sites are available for analysis, #though the original paper reports testing ~50,000 sites. Why is there such a #discrepancy? -#original paper excludes 15 of 398 patients for analysis, yet there are still -#394 patients included in the dataframe. Why are there not 385 patients -#included, as reported in the paper? +#Also, the original paper excludes 15 of 398 patients for analysis, yet there +#are still 394 patients included in the dataframe. Why are there not 385 +#patients included, as reported in the paper? ### Some contrasts @@ -165,19 +165,34 @@ fit3 <- eBayes(lmFit(exprs(DNAme)[, as.integer(rownames(design3))], design3)) ``` #Paige's code review: - ##The following additions were necessary in the first chunk of code: - #1) BiocManager::install("GEOquery") - #2) BiocManager::install("limma") + 1)The following additions were necessary in the first chunk of code: + A) BiocManager::install("GEOquery") + B) BiocManager::install("limma") +2)Comments on quality of code logic, accompanying text, labels: + A) Though I am unfamiliar with some of the functions used in the current + code, I was able to follow the logic decently well. + B) I would have made changes to the text accompanying the code output in + the section, "Do we see genome-wide hypermethylation from TET2 + mutations?" to indicate that the number of probes corresponds to the + number of CpG sites hypermethylated in samples harboring a specific + type of mutation. + C) Regarding labels: I would have preferred no labels on the x-axis of + the heatmap; it was rather distracting trying to figure out what the + small, inscrutable black text meant. #Paige's science review: 1) Assumptions regarding the term "probe": - A) I assume a "probe" is one of the ~50,000 CpG islands sites + A) I assume a "probe" is one of the ~50,000 CpG sites that the authors looked at in the paper. B) I also assume that "6513 probes for IDH" (for example) infers that the same 6513 CpG islands are mutated across all samples harboring an IDH mutation. - 2) According to the heatmap provided (under #always plot your data), - it is evident IF either IDH1/2 or TET2 mutations are present in the AML + i) Should (B) be the case, indicating the same 6513 sites are + preferentially hypermethylated in AML patients harboring IDH + mutations should be indicated in the text accompanying the code. + 2) Mutually exclusive TET2 and IDH1/2 mutations: + According to the heatmap provided (under #always plot your data), + it is evident that IF either IDH1/2 or TET2 mutations are present in the AML patients, these mutations LARGELY occur mutually exclusive of each other. However, it should be noted that most of the samples harbored neither mutation. Additionally,the heatmap also demonstrates a small population of samples that @@ -186,10 +201,17 @@ fit3 <- eBayes(lmFit(exprs(DNAme)[, as.integer(rownames(design3))], design3)) A) Question: If the vast majority of the samples extracted from these AML patients do not harbor either mutation, is this really a defining feature of AML? - 3) The 6500+ hits for methylated CpG sites in IDH mutated-samples vs. + 3) Hypermethylation profiles and IDH and TET2 mutations: + The 6500+ hits for methylated CpG sites in IDH mutated-samples vs. the ~10 hits for the TET2 mutated-samples indicates that these mutations have differential associations with hypermethylation at CpG sites. It is misleading for the authors to indicate that both kinds of mutations engender "overlapping" hypermethylation signatures, considering that even if all TET2-related sites overlapped with the IDH-related sites, this only represents 0.15% of all - sites implicated by IDH mutation. \ No newline at end of file + sites implicated by IDH mutation. Also, hypermethylation at ~10 CpG spots (in + the context of TET2 mutation) is not sufficient to constitute a + genome-wide hypermethylation profile.Therefore, it seems IDH mutation has a + larger association with hypermethylation profiles, while TET2 mutation has a + minimal association with hypermethylation profiles. + + \ No newline at end of file