Hello!
I have received some custom vcfs (yes, I know, and I hate it) containing information of each sample per row, independent of variant. This means that some variants are repeated.
I have the following columns:
columns 1-17
#CHROM POS GENE REF ALT QUAL FILTER INFO PTID AF AQ GT DP AD AB GQ PL
GT is genotype, AF is allele frequency, DP is depth and AB is the fraction of each allele. I don`t care for extra information except those that do define the frequency or genotype of alleles.
So for some variants:
PTID
VARIANT A SAMPLE1
VARIANT A SAMPLE2
etc
Where Filter, Qual and Info are empty. PTID is the ID for each sample
What I need is a multisample vcf like
#CHROM POS GENE REF ALT QUAL FILTER INFO SAMPLE1_genotype SAMPLE2_genotype etc
Is there any way of converting this to a multi sample vcf based on the PTID? I have no idea where to start as each variant is defined by pos gene ref and alt (I assume), and I`m worried of messing with the file and getting something wrong without knowing
It would be easier to just have the normal VCF, but the institution that gave this custom VCF is not easily accessible
use awk . Something like this ( I'm too lazy to understand and retrieve all the fields..)
awk 'BEGIN{printf("##INSERT VCF HEADER HERE\n#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tSAMPLE\n");} {printf("%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\tPTID=AF=%s;AQ=s;GENE=\n",$1,$3,$4,$5,$6,$7,$8,$9,$10,$11,$2);}' input.txt
I think this might be getting closer, but still I'm not sure how to make the genotype for each sample into a new column like a normal multisample vcf. I've updated the question to make this more clear
I had something similar while back: Converting TSV file to VCF
It would be easier to just have the normal VCF, but the institution that gave this custom VCF is not easily accessible
VCF files must adhere to the VCF File Format Specifications. The files you have been given are not VCF files. A "Custom VCF" would be a VCF file that includes INFO and/or FORMAT fields that are not defined in the specifications (e.g. "GT" is specifications-defined but "RANDOMinfoFIELDname" is not).
It would be easier to just have the normal VCF
I would ask your data provider for the actual VCF files for this data. What is likely to have happened is that they've generated real VCF files, done some data transformation on them, and you've been given the custom TSV file output from those transformations. Those TSVs may be useful for the some of the analysis they were doing but they're clearly not useful to you as, not being VCF files, you can't use any of the many many tools that input/output VCF.
I agree with everything. Because of internal policies this partner will not give the full VCF, just this TSV. I could transform it all back to a VCF format because all data was there, but I had to make some transformations and operations to the data!
I ended up using R for this question. The R community is very active and helped reach to an answer (which I customized).
#df.list is your list of dataframes/custom vcfs. I imported them as a named list.
for (i in seq_along(df.list)) {
olddata_long = df.list[[i]]
#get the correct columns of your vcf file. Mine were 1-17
olddata_long = olddata_long[1:17]
#Create FORMAT COLLUMN according to VCF format. GT should be the first
olddata_long["FORMAT"] = "GT:AQ:DP:AD:AB:GQ:PL"
#Correct column orders (FORMAT must be before sample names. GT must be the first after. Columns should be in the order you wrote down on format
olddata_long = olddata_long[c(1:8, 18, 9,12,11,13:17)]
#transform to data wide, separated by : (vcf format standard)
data_wide = olddata_long |>
mutate(across(-c(1:10), ~ paste0(.x))) |>
unite("value", -c(1:10), sep = ":") |>
pivot_wider(names_from = PTID, values_from = value)
#change NA to ./.:.:0:.:0,0:.:. . In my case, I had to customize the NA to match the NA values of each specific column. For example, GT NA value is ./. NA values are separated by :
data_wide[is.na(data_wide)] = "./.:.:0:.:0,0:.:."
#return dataframe to list
df.list[[i]] = data_wide
}
I`m not 100% sure that this is in the correct complete VCF format now, but I can annotate with VEP, use BCFtools and tabix. Annotation is working fine on it, so I think its OK.
I can split my files before hand if they are too big! I`m sure there might be ways that uses less memory, but this works.
Yeah I guess the presence of samples as rows makes a pivot-wider operation necessary, which makes awk sub-optimal. Good call.
It does seem like the previous answers address a now-altered question, so please don't edit your question in response to an answer when it makes the answer irrelevant. Instead of removing existing content, add corrections/changes as notes so people can follow the evolution of a post. In conjunction with the lack of follow up on your other posts (as listed in a comment above), this starts to look like you are abusing the forum, so please be mindful of your actions.
Log in to answer this question.
You have many questions without comments or validation:
I'm not sure what this means, but most questions also did not have many answers. I always give a thumbs up when the answers works so I'm not sure what is the expectation here! Didn't find any very informative description on the forum rules also
You should accept the answers that solve your question and/or follow-up others comments and advices with your experience for anyone come across your posts in the future. These posts come up in the search engines so someone else with the same problem would have easier time figuring out what worked and didn't work for you.