Kallisto "pseudo" output to counts/expression matrix

Question

[paulranum11]

Is there a tool available that will convert the output of kallisto "pseudo" (matrix.tsv, matrix.ec, and matrix.cells) into a standard genes by cells counts matrix?

[Kevin Blighe]

May be something - could contact the Pachter lab directly. They are not regulars here on Biostars.

Other options:

• import into DESeq2 using tximport (you'll just need to make a file listing, and then tximport will do most of the heavy lifting)
• paste relevant columns together yourself into a matrix

[paulranum11]

The output of kallisto "pseudo" is formatted differently than the output of other kallisto runs i have performed in the past (kallisto "quant" etc). The output appears to be associative arrays (matrix.tsv, matrix.ec, and matrix.cells). Here are some examples:

matrix.tsv

12531   0   1
18862   0   1
384133  0   1
138357  1   1
199880  1   1
422948  1   1
51111   2   1

matrix.ex

422968  22108,80043,80044,84616,91003,93151,93152,93153,93154
422969  939,941,950,974,1125,4632,102217,102409,113753,113766,115246,115359
422970  102268,102280,108561,110740,114873
422971  91677,91680
422972  66108,66109,76096
422973  8687,8688,39455
422974  15300,18434,40489,94720
422975  58132,103245,104106,105656,106906,109217,109621,109756,109846,109853,109855,110441,111708,112144
422976  6117,6118,80071,90398
422977  19697,60885,68905

matrix.cells

cell_001
cell_002
cell_003
cell_004
cell_005

I took a look at using tximport but it appeared to me to be better suited for the output of kallisto "quant" it was not clear where i would begin with the output of kallisto "pseudo". Are there any examples availible for running tximport with this style of input?

[Kevin Blighe]

I would contact the Pachter group.

Answer 1

[h.mon]

The R/BioConductor tximport can do this.

[paulranum11]

tximport does not support this kallisto "pseudo" format output.

[h.mon]

There is a suggestion to use python at the kallisto-sleuth-users list:

kallisto pseudo output