chanzuckerberg / expressionmatrix2 Goto Github PK

Top differentially expressed genes for a cell.
Relative to what?

Python-callable function to remove a cell set.

Implemented as removeCellSet.

I implemented this as follows:
createCellSetUsingNumericMetaDataGreaterThan(cellSetName, metaDataName, lowerBound)
createCellSetUsingNumericMetaDataLessThan(cellSetName, metaDataName, upperBound)
createCellSetUsingNumericMetaDataBetween(cellSetName, metaDataName, lowerBound, upperBound)

Charlotte, please close after verifying that it satisfied your needs.

When an ExpressionMatrix object is created from an existing directory with allowReadOnly=true, a write attempt on read-only data results in a segmentation fault. This causes the calling Python script to terminate immediately, without cleanup.

getCellsExpressionCountFromGenes(cellIds, geneIds) - just one step further to get the whole matrix for a list of cells and genes.

Hi Paolo, I was wondering if there is support for using python 3.6? Currently, I run into this issue when I try to run "python run.py"

Traceback (most recent call last):
File "ExpressionMatrix2-0.4.0/tests/ToyTest1/run.py", line 4, in
import ExpressionMatrix2
ImportError: Python version mismatch: module was compiled for version 3.5, while the interpreter is running version 3.6.

Hi Paolo,
I'm trying to use your package on Mac OS X by manually installing the dependencies via conda. I realize this isn't exactly following your instructions but I wanted to try anyway :) If this works, then we can make a portable package that anyone on any OS can use, not just Ubuntu, since conda will handle the C/C++ dependencies for us.

I was able to install boost and Python and Graphviz. Here is the full log of my installation. The punchline is that when I try to make all, I get an error that seems to have to do with locating boost libraries:

(ExpressionMatrix2) ➜  Release git:(master) make all
Building file: ../src/CellSets.cpp
Invoking: GCC C++ Compiler
g++ -std=c++0x -I/usr/include/python3.5m -O3 -msse4.2 -Wall -Wconversion -c -fmessage-length=0 -fPIC -MMD -MP -MF"src/CellSets.d" -MT"src/CellSets.d" -o "src/CellSets.o" "../src/CellSets.cpp"
In file included from ../src/CellSets.cpp:1:
In file included from ../src/CellSets.hpp:5:
In file included from ../src/MemoryMappedVector.hpp:7:
../src/CZI_ASSERT.hpp:8:10: fatal error: 'boost/lexical_cast.hpp' file not found
#include <boost/lexical_cast.hpp>
         ^
1 error generated.
make: *** [src/CellSets.o] Error 1

Do you have any suggestions?
Thanks!
Olga

Document default values for ClusterGraphParameters.

Done.

Top differentially expressed genes for a cluster
Differential expression of all genes of cluster(s) vs other cluster(s)

Implemented as getClusterGraphGenes and getClusterAverageExpression.

Charlotte, please close if this satisfies your needs, or else reassign to me with comment.

Dear authors,
This is a very great job. Is there any publicly published paper to understand the process of clustering with LSH in more detail? Thanks a lot.

Get a list of clusters for a given graph: implemented as getClusterGraphVertices.
Get cell id list for each cluster: implemented as getClusterCells.
Use createClusterGraph to create one or more cluster graphs for each cell graph.

Charlotte, please close after verifying that this satisfies your needs, otherwise reassign to me with comment.

In the getCellExpressionCount family of functions, add an option to eliminate output for genes where there is no expression across the whole set of cell ids.

Be able to send several filters at once instead of doing each at a time and then doing the intersection.

For categorical queries, this capability is provided to some extent by the syntax for regular expressions.

Other complex queries can be implemented one step at a time, so this is an efficiency issue, not a functionality issue. Therefore postpone for now.

Otherwise regex queries coming from the browser have to be escaped.
Example:
s = e.createCellSetUsingMetaData("s3{}".format(run), "Selection", "Astrocytes(HEPACAM)") → 0 matches
s = e.createCellSetUsingMetaData("s3{}".format(run), "Selection", "Astrocytes(HEPACAM)") → 714 matches