tbj128 / mian Goto Github PK

Hi,

The install instructions in INSTALL.md are outdated.
More specifically, any modern Linux distribution has python >= 3.9

It might be possible to create a conda environment with python 3.6.7 as is required now. Personally, I could not get it to work.

Regards

Taxonomy information is currently limited to the taxonomic levels (eg. phylum, class, order, etc). Some use cases may have additional taxonomic groupings, such as "OTU guilds" which can be an area of investigative interest.

Hi,
Recently i tested Mian for my data. A few things which I think might be very helpful and if possible can be integrated into future versions:

1. Regarding data processing: Currently, Mian has sub-sampling and lacks other filters like normalization, log transformation, rarefaction option.
2. An important feature while visualization is filtering n number of top (down) taxa to make te visualization process more easy.
3. More export options like vector image (svg,eps) or higher resolution scalar images would greatly enhance Mian.

Hope, this might be possible in future versions of Mian. Thanks for the great tool and all the hard work went into Mian.

When viewing data with composition barplot the "top 10 labels" get cut off at phylum level.

Request to add a total species accumulation curve for all samples in the data set. Total species on the y and total number of reads on the x.

When creating a boxplot that is filtered by metadata, the groups of metadata that are excluded still appear in the legend and x-axis. They will always be zero, so it would only make sense to not plot them.

Also, the alignment is off on the x-axis.

Mian currently supports ad-hoc filtering which can remove or include certain samples and/or OTUs at the time that the visualization or feature selection tool is run. This is done by design to enable users to quickly 'zoom-in' into certain features of interest or to temporarily exclude outliers.

There may be merit to perform a 'one-time' filtering when the data is being uploaded. This way, users can immediately exclude some samples (eg. perhaps from a mock community) and apply some subsampling over the filtered data set.

Test