Hi, I am following a course on bulk RNA-seq and am currently on the lecture covering index construction using Kallisto. I built my index using the latest Ensembl release (Release 115: homo_sapiens/cdna) with Kallisto v0.51.0, whereas the instructor used Release 108 with Kallisto v0.48.0. My resulting index file is only 550 MB, while the instructor's file is 2.6 GB. Could the combination of the newer Ensembl release and the update to Kallisto v0.51.0 explain such a significant discrepancy in size? I have verified that my cDNA file contains 328,868 transcripts using grep, so the sequences are definitely present. I just wanted to confirm if this size discrepancy is expected with the v0.51.0 update.
•
written
by
musfirajamil •
2 answers
https://github.com/pachterlab/kallisto/releases
This will tell you that a new index structure was introduced in v.0.50.0 which probably explains what you see.
The current maintainer/developer dsull is here, I am sure he can answer detail questions if you have them.
Thanks ATpoint for the mention and he's 100% correct. The index implemented in kallisto versions >0.50 is a much more compact index. The results of mapping your reads, of course, will be identical to those of previous versions if you build the index the same way.
(P.S. There are other ways to build the index in the new version that improve accuracy for single-cell data, but that's for another day; also, the reason we made a more compact index in the new version was so that memory wouldn't explode if one wanted to index all the introns in the human genome, e.g. for single-nucleus data).
Log in to answer this question.
-
Building Salmon and STAR index using rat GRCr8 reference genome
written by Ajina •
-
indexing the drosophila genome for kallisto
written by gogeni5529 •
-
kallisto index build difference according to version
written by estilo •
-
Do indexes made with cDNA fasta file contain only mRNAs?
written by bioinfo •
-
How do I get estimated gene counts for a reference that contains GFP using Kallisto?
written by bioinfo •
-
How to filter .fa file by gene_biotype
written by bioinfo •
-
Building a new index for HISAT2 for RNA-Seq data analysis
written by Archit •
-
STAR index/alignment error : no exons found
written by sandeep.amberkar18
-
what is the corresponding proteome or transcriptome (cDNA fasta in Ensembl) file in NCBI?
written by SMILE
-
Should kallisto take hours to run?
written by Kristin Muench
As long as the index files are working fine don't worry about the file sizes. File sizes should never be used for diagnostic purposes (especially across different versions of the program and/or operating systems).