poisonalien / tcgamutations Goto Github PK

Hi, thank you so much for all your powerful tools!

I was thinking to compare my data with the tcga precompiled mafs in terms of visualizing the mutations on certain genes (aka lolliplots for two maf files). Problem is that due to space there is no protein change or AA change column. And when I use the HGVSp_Short I get an error (Error in subsetMaf(maf = maf, includeSyn = FALSE, genes = geneID, query = "Variant_Type != 'CNV'") :
trying to get slot "maf.silent" from an object (class "data.table") that is not an S4 object ).

Do you think there is a faster way to do that through maftools?
Was thinking to annotate the maf but it doesn't really work...I also tried to extract only chr, start, end and make a new maf for the genes that I want ...so that is why I am reaching out to you in case you have a "quicker" way to do this.

best
Iris

Hi.
I have a following error message.
Error: object 'tcga_laml_mc3' not found

Do I need to download tcga_laml_mc3 file from somewhere?

Thanks.

Hi,
i am using the below commands:
icgc =system.file("extdata", "ssm_orca_In.tsv.gz", package="maftools")
maf <- icgcSimpleMutationToMAF(icgc = icgc, addHugoSymbol = TRUE)
Converting Ensemble Gene IDs into HGNC gene symbols.
Done! Original ensemble gene IDs are preserved under field name ens_id
--Removed 904661 duplicated variants
--Found 101696 variants with no Gene Symbols
--Annotating them as 'Unknown' for convenience
--Non MAF specific values in Variant_Classification column:
NA
--Non MAF specific values in Variant_Type column:
NA

I am getting the below error:
plotmafSummary(maf)
Error in (function (classes, fdef, mtable) :
unable to find an inherited method for function ‘getSampleSummary’ for signature ‘"data.table"’

Hi,

Just wondering about feasibility of adding additional sample metadata, for example which capture kit was used?

Reason we care about capture kit:

Capture kit biases have led to false negative mutation calls in multiple TCGA cohorts. Variability in the capture kits used can contribute to inter-cohort TMB variation. More importantly, there are some genes that may be mutated in a sample, but whose mutation you will/won't see because of the specific capture kit that was used. See Wang et al. 2018 for details

So I am new to R, new to Github, new to coding... I apologize for simple minded questions. I am trying to generate a gene expression analysis heatmap using 10 genes (10 genes). I basically want to see the number of mutations in these genes in each of the TCGA cancers. I was directed to install the maf file from the https://gdc.cancer.gov/about-data/publications/mc3-2017 website. I am also using a Mac. So I downloaded the file "mc3.v0.2.8.PUBLIC.maf" and everytime I download it, I open to see how big the file is and its only 66KB on disk! Would you be able to help me?

@PoisonAlien It is strange that there are many duplicated variants if I recreate the MAF, why this happen?

> tcga_load("luad")
Loading objects:
  tcga_luad_mc3
Successfully loaded TCGA LUAD!
Please cite https://doi.org/10.1016/j.cels.2018.03.002 for MAF source.
> maftools::read.maf(rbind(tcga_luad_mc3@data, tcga_luad_mc3@maf.silent))
-Validating
--Removed 18834 duplicated variants
-Silent variants: 63940 
-Summarizing
--Possible FLAGS among top ten genes:
  TTN
  MUC16
  USH2A
  FLG
-Processing clinical data
--Missing clinical data
-Finished in 5.521s elapsed (5.024s cpu) 
An object of class  MAF 
                        ID summary    Mean Median
 1:             NCBI_Build      NA      NA     NA
 2:                 Center      NA      NA     NA
 3:                Samples     515      NA     NA
 4:                 nGenes   17130      NA     NA
 5:        Frame_Shift_Del    3675   7.150    5.0
 6:        Frame_Shift_Ins    1096   2.132    1.0
 7:           In_Frame_Del     360   0.700    0.0
 8:           In_Frame_Ins      33   0.064    0.0
 9:      Missense_Mutation  122329 237.994  165.5
10:      Nonsense_Mutation   10152  19.751   12.5
11:       Nonstop_Mutation     164   0.319    0.0
12:            Splice_Site    4094   7.965    5.0
13: Translation_Start_Site     206   0.401    0.0
14:                  total  142109 276.477  189.0

Hi,
I am getting really different mutational frequencies in some genes, when looking at data from the Broad Firehose, compared to the MC3 source. I would expect some differences due to the documented differences in preprocessing, but this effect is quite extreme in some cases. For example, looking at the top 10 mutated genes in colon cancer for Broad and MC3:

Broad

MC3
If you look at the PTEN gene for example, this is mutated in over 40% of subjects in the Broad firehose data. This is way above its mutation rate in this cancer from the literature, including in the TCGA marker paper. Further, this gene is mutated in only 3% of cases in the MC3 data. Do you have an idea where such a large discrepancy could come from? Help here much appreciated!

Hi PoisonAlien, I am very very new to bioinformatics and I apologize if this is a dumb questions. Would it be possible to show how you can apply your tools to CCLE maf file? I tried separating the file to individual lineage maf files but having difficulty in running MutsigCV on them. I am sorry if I am not making sense. My question is that if you can apply your tool for CCLE maf file.

Thanks.

Shwetha

LUSC Firehose dataset includes two samples with only 1 mutation each (both silent)

This makes these samples suspicious enough that they may warrant being filtered out from the RDS. They're still present in maf.silent slot, which is not very problematic, however they get added to the variant.classification.summary slot when merged with merge_mafs

See issue for details:
PoisonAlien/maftools#904

Hello,
I am very interested in comparing my data with the TCGA mutations. My data (mutations in RNA-seq data)is in VCF file format which I converted to MAF files. I wanted to check with if you if the available datasets in your package are from whole exome sequencing data ? Do you have any datasets that are mutations from TGGA RNA-seq cohorts ?
Thanks, K

Hi!
After loading the study
How could I use them in maftool by reading maf?

tcga_load("luad")
Loading LUAD. Please cite: https://doi.org/10.1016/j.cels.2018.03.002 for reference
An object of class MAF
ID summary Mean Median
1: NCBI_Build NA NA NA
2: Center NA NA NA
3: Samples 517 NA NA
4: nGenes 17130 NA NA
5: Frame_Shift_Del 4021 7.778 5
6: Frame_Shift_Ins 1185 2.292 1
7: In_Frame_Del 388 0.750 0
8: In_Frame_Ins 37 0.072 0
9: Missense_Mutation 133671 258.551 177
10: Nonsense_Mutation 11074 21.420 13
11: Nonstop_Mutation 179 0.346 0
12: Splice_Site 4469 8.644 5
13: Translation_Start_Site 225 0.435 0
14: total 155249 300.288 202

I tired...

maftools::read.maf(rbind(tcga_luad@data))
Error in rbind(tcga_luad@data) : object 'tcga_luad' not found

or

luad = read.maf(maf = 'inst/extdata/MC3/luad.RDs')
-Reading
Error in data.table::fread(file = maf, sep = "\t", stringsAsFactors = FALSE, :
File 'inst/extdata/MC3/luad.RDs' does not exist or is non-readable. getwd()=='/Users/dingdingdingplus'

Thank you very much!

Error in mafSurvival(maf = tcga_stad, genes = "TP53", time = "days_to_last_followup", Status = 'vital_status', isTCGA = TRUE):

Overall_Survival_Status not found in clinicalData. Use argument Status to povide column name containing events (Dead or Alive).

error msg told that vital status should be assigned values dead/alive, while in the package assigned binary value 0/1.

How should i fix it?

I was using TCGAmutations to obtain LUAD mutation and clinical data from 'TCGA MC3', so I can plot survival data (Kaplan–Meier) using the mafsurvival function of the maftools library.

I noticed differences between the mafsurvival generated graph (LUAD for the CDKN2A gene) with the survival graph generated on-line (https://www.tcga-survival.com/data-table?view=gene&gene=CDKN2A&filter=cancer_type&cancer_type=LUAD).

To determine the cause of this, I then downloaded the mc3 MAF and the clinical data from "https://gdc.cancer.gov/node/905/":

1. mc3.v0.2.8.PUBLIC.maf
2. clinical_PANCAN_patient_with_followup.tsv
   I then processed the files with maftools after subsetting for LUAD samples.
   When I performed 'mafsurvival' for the CDKN2A gene with this data, it was identical to the www.tcga-survival.com website.

To determine the discrepancies, I compared the specific clinical data downloaded by TCGAmutations and the ones downloaded from "https://gdc.cancer.gov/node/905/".

Here are a few examples of the inaccurate info I found from the pre-compiled TCGAmutations clinical data:
TCGA-44-5645-01A-51D-A27T-08 , days to last followup=208, (should be 852)
TCGA-44-7662-01A-11D-2063-08 , days to last followup=50, (should be 218)
TCGA-53-A4EZ-01A-12D-A24P-08 , days to last followup=280, (should be 1071)
TCGA-55-8089-01A-11D-2238-08 , vital status = Alive ; days to death=NA, (should be dead and 702, respectively)
TCGA-73-A9RS-01A-11D-A410-08 , vital status = Alive ; days to death=NA, (should be dead and 340, respectively)

I also went to "https://portal.gdc.cancer.gov/" to search for these specific cases and confirmed that the pre-compiled TCGAmutations data was inaccurate.

Maybe I'm not processing the clinical data from TCGAmutations correctly:

library(TCGAmutations)
library(maftools)
tcga_luad <- tcga_load(study = "LUAD")
luad_maf = read.maf(maf = tcga_luad@data, clinicalData = [email protected])

Hello!

I am having some issues in passing samples' clinical data to use it as mentioned in maftools vignettes. I have tried a desperate:

cancer.maf <- toga_acc_mc3
cancer.clinical <- getClinicalData(toga_acc_mc3)

cancer <- read.maf(maf = cancer.maf, clinicalData = cancer.clinical)

and gives an error:

Error in (function (classes, fdef, mtable) :
unable to find an inherited method for function 'getClinicalData' for signature '"character"'

However, reading the papers for TCGAmutations and Maftools I couldn't figure out how could be possible to connect one with the other.

May you please help me?

Thank you