I strongly recommend against downloading files from non-official repositories. Always download from the official sources like UCSC, GENCODE, etc. You have no control over what the uploader might have done with the file. Not saying it was not proper, it is just not guaranteed.
I am sure this may have been answered already, Can some one direct me to a bed file of TSS of hg19 or how can be extract / clean it. I googled it but could not find it.
4 answers
Go to UCSC table browser and select hg19 gene models from either ensembl/Refseq/Gencode, which is either a GTF or BED file.
If gene is on + strand, col-2 is TSS and if gene is on -ve strand, col3 is TSS.
human refGene hg19 TSS:
https://raw.githubusercontent.com/Shicheng-Guo/AnnotationDatabase/master/hg19/refGene_hg19_TSS.bed
human refGene hg38 TSS:
https://raw.githubusercontent.com/Shicheng-Guo/AnnotationDatabase/master/hg38/refGene.hg38.TSS.bed
How to get human refGene hg38 TSS?
Try this one:
https://raw.githubusercontent.com/Shicheng-Guo/AnnotationDatabase/master/hg38/refGene.hg38.TSS.bed
Download the GTF from their official repository at NCBI, then filter for entries of the type "transcript" and take the start coordinate for plus-strand features and end coordinate for bottom strand features. That's it.
For refGene_hg19_TSS.bed file, the TSS site of "-" strnad should subtract 1, because the bed file not contain the last coordinate.
Log in to answer this question.
This post can guide you: https://randomstate.net/2018-06-28-getting-refseq-gene-tss-from-ucsc/