It's easily solved by just doing a "pip install anndata", but might confuse non-techies.

It appears that pandas eats the first row when building the dataframe. I'm not very familiar with pandas, but I believe that's normal behavior, and how it gets the dataframe's column names.

On output, the first row is a series of integers that I'm guessing is an index into part of the dataframe's internal structure. The second row is the first column of the input. The first column of the output is actually the second row of the input.

I circumvented this with
head -1 input_file >heading_file cat heading_file input_file >temp_file

Then I did a transpose_csv on temp_file, and got the first column right. Doing a tail +1 on the transposed temp_file should get rid of the "mystery" row and produce the desired result.

If you're on Linux or MacOS, you should be able to do a little piping to automate this. If you're on Windoze ... well, you shouldn't be ;-)

Thanks for doing big-csv: I didn't know enough about pandas to know it could be used, and though I was going to have to write my own transposition code. It's a real time saver.

I'm guessing that the name was changed at some point, but the documentation wasn't updated.