profile
viewpoint
Alex Rogozhnikov arogozhnikov Aperture Science https://arogozhnikov.github.io ML researcher

arogozhnikov/python3_with_pleasure 3636

A short guide on features of Python 3 with examples (updated for python 3.7)

arogozhnikov/einops 805

Deep learning operations reinvented (for pytorch, tensorflow, chainer, gluon and others)

arogozhnikov/infiniteboost 177

InfiniteBoost: building infinite ensembles with gradient descent

arogozhnikov/3d_nn 137

Raymarching neural network in browsers with WebGL shaders

arogozhnikov/hep_ml 124

Machine Learning for High Energy Physics.

arogozhnikov/arogozhnikov.github.io 95

'Brilliantly wrong' blog, Machine Learning visualizations live here

anaderi/lhcb_trigger_ml 15

LHCb trigger based on machine learning research

arogozhnikov/cats 8

Test of algorithms on highly-categorical data

arogozhnikov/DeepMMD-GAN 8

Yet another (very simple) approach for adversarial training.

arogozhnikov/OBDT 4

Pruning experiments

pull request commentopencv/opencv

RFC: Jpeg2000 OpenJPEG port

I see this is very recent change (only 4 days ago), but I'm interested in testing openjpeg for my case (uint16 grayscale) to read/write from python. When should I expect this functionality available in distribution?

StefanBruens

comment created time in 7 days

issue commentsamtools/bcftools

compilation error

@pd3 you're correct, I got some older htslib. Now it builds successfully, thanks!

aderzelle

comment created time in 13 days

issue commentsamtools/bcftools

compilation error

@pd3 just tried, did not worked. Full output

make clean && ./configure && make
rm -f test/*.o test/*~ test/test-rbuf test/test-regidx
rm -f test/*.hex
rm -f plugins/*.so plugins/*.P
rm -rf plugins/*.dSYM
rm -f gmon.out *.o *.a *~ bcftools version.h
rm -rf *.dSYM test/*.dSYM
checking for gcc... gcc
checking whether the C compiler works... yes
checking for C compiler default output file name... a.out
checking for suffix of executables...
checking whether we are cross compiling... no
checking for suffix of object files... o
checking whether we are using the GNU C compiler... yes
checking whether gcc accepts -g... yes
checking for gcc option to accept ISO C89... none needed
checking for grep that handles long lines and -e... /usr/bin/grep
checking for C compiler warning flags... -Wall
checking for special C compiler options needed for large files... no
checking for _FILE_OFFSET_BITS value needed for large files... no
checking location of HTSlib source tree... ../htslib
checking shared library type... Darwin dylib
checking for library containing dlopen... none required
checking if the compiler accepts -rdynamic... yes
checking for zlib.h... yes
checking for inflate in -lz... yes
checking for library containing regcomp... none required
configure: creating ./config.status
config.status: creating config.mk
config.status: creating config.h
config.status: config.h is unchanged
echo '#define BCFTOOLS_VERSION "1.10"' > version.h
gcc -Wall -g -O2 -I. -I../htslib   -c -o main.o main.c
gcc -Wall -g -O2 -I. -I../htslib   -c -o vcfindex.o vcfindex.c
gcc -Wall -g -O2 -I. -I../htslib   -c -o tabix.o tabix.c
gcc -Wall -g -O2 -I. -I../htslib   -c -o vcfstats.o vcfstats.c
gcc -Wall -g -O2 -I. -I../htslib   -c -o vcfisec.o vcfisec.c
gcc -Wall -g -O2 -I. -I../htslib   -c -o vcfmerge.o vcfmerge.c
gcc -Wall -g -O2 -I. -I../htslib   -c -o vcfquery.o vcfquery.c
gcc -Wall -g -O2 -I. -I../htslib   -c -o vcffilter.o vcffilter.c
gcc -Wall -g -O2 -I. -I../htslib    -c -o filter.o filter.c
filter.c:2902:86: error: use of undeclared identifier 'VCF_OVERLAP'
            else if ( !strcasecmp(out[ival].key,"overlap") ) { out[ival].threshold = VCF_OVERLAP<<1; out[ival].is_str = 0; }
                                                                                     ^
1 error generated.
make: *** [filter.o] Error 1
  • have you tried xcode? No, I did not (and I did not need it anywhere else)

Perl itself is installed with brew.

Don't see anything about arch in config.mk. Full paste:


prefix       = /usr/local
exec_prefix  = ${prefix}
bindir       = ${exec_prefix}/bin
libexecdir   = ${exec_prefix}/libexec
datarootdir  = ${prefix}/share
mandir       = ${datarootdir}/man

CC       = gcc
CPPFLAGS =
CFLAGS   =  -Wall -g -O2
LDFLAGS  =
LIBS     =

DYNAMIC_FLAGS = -rdynamic

USE_GPL =
GSL_LIBS =
PERL_CFLAGS =
PERL_LIBS =

PLATFORM   = Darwin
PLUGINS_ENABLED = yes
plugindir  = $(libexecdir)/bcftools
pluginpath = $(libexecdir)/bcftools
PLUGIN_EXT = .so

HTSDIR = ../htslib
include $(HTSDIR)/htslib.mk
include $(HTSDIR)/htslib_static.mk
HTSLIB = $(HTSDIR)/libhts.a
HTSLIB_LIB = $(HTSLIB) $(HTSLIB_static_LIBS)
HTSLIB_DLL = $(HTSDIR)/
HTSLIB_LDFLAGS = $(HTSLIB_static_LDFLAGS)
W32_PLUGIN_LIBS = libbcftools.a $(HTSLIB_DLL) $(ALL_LIBS)
BGZIP = $(HTSDIR)/bgzip
TABIX = $(HTSDIR)/tabix
HTSLIB_CPPFLAGS = -I../htslib
#HTSLIB_LDFLAGS = -L../htslib
#HTSLIB_LIB = -lhts
#W32_PLUGIN_LIBS = libbcftools.a $(HTSLIB_LDFLAGS) $(HTSLIB_LIB) $(ALL_LIBS)

Probably that's all I can help with

aderzelle

comment created time in 14 days

issue commentpysam-developers/pysam

Support for serialization?

+1 for this issue. So far my best shot is to convert to dictionary. I see no other way to parallelize and aggregate reads.

llevar

comment created time in 19 days

issue commentpysam-developers/pysam

iterate ONLY over the unmapped reads

In my case all unmapped alignments are not assigned to any contig. Still no better way than just iterate over whole file?

smangul1

comment created time in 21 days

issue commentpysam-developers/pysam

Reset the iterator fetch() returned.

double this issue. So far the only way I know is to open the file again

yangyxt

comment created time in 21 days

issue commentsamtools/bcftools

compilation error

facing same issue on mac os. Configure does not complain. Disabling libperl filters does not help.

Is there a recommended way I should get libperl on mac? Brew does not provide it

aderzelle

comment created time in a month

pull request commentarogozhnikov/hep_ml

Pandas "ix" indexer removed.

Thanks for reminding @itsayushthada . .ix was used in several other places, I got it all fixed, new version is on pypi, so you can upgrade with pip install hep_ml==0.6.1

itsayushthada

comment created time in a month

push eventarogozhnikov/hep_ml

Alex Rogozhnikov

commit sha 114ac9e896c3a601761092760a7b315f448d59c6

fix layout

view details

push time in a month

push eventarogozhnikov/hep_ml

Alex Rogozhnikov

commit sha 718c7d250173fa5c64398fd0f8e4e37dba8c3f2a

remove appveyor (it stopped supporting previous format)

view details

Alex Rogozhnikov

commit sha 39dd942d6333576e898243c267f2d89f9e738d95

fix tests

view details

push time in a month

push eventarogozhnikov/hep_ml

Alex Rogozhnikov

commit sha 4ad4534e9e6b3f528b14953b32ca4a9ed521735a

bump version

view details

push time in a month

push eventarogozhnikov/hep_ml

Alex Rogozhnikov

commit sha 48c5f4ab57f076113c56475f3cd87b4bdfa6f4ab

updates to eliminate usage of .ix from pandas

view details

push time in a month

push eventarogozhnikov/hep_ml

Ayush Thada

commit sha 15f3f4757cf0fddb32853e83ace75af55d054bd3

Pandas "ix" indexer removed. New indexing added. "ix" indexer was last supporting in Pandas 0.22.0.

view details

push time in a month

PR merged arogozhnikov/hep_ml

Pandas "ix" indexer removed.

New indexing added. "ix" indexer was last supported in Pandas 0.25.3. Tested the new code in Pandas 1.0.0 and Pandas 1.0.1.

+1 -1

0 comment

1 changed file

itsayushthada

pr closed time in a month

PR closed statgen/demuxlet

Add option to produce only singlets probabilities (fix #52)

Proposing PR because running demuxlet against large number of genotypes drives to OOM

+6 -0

3 comments

1 changed file

arogozhnikov

pr closed time in a month

issue commentsaic-vul/fbrs_interactive_segmentation

Some controls are missing on mac os in GUI (?)

@ptrvilya thanks for response! I use brew-provided python, but still that looks like most probable reason (since tkinter says it uses 8.5). Unfortunately can't verify your guess in details.

arogozhnikov

comment created time in 2 months

issue openedsaic-vul/fbrs_interactive_segmentation

Some controls are missing on mac os in GUI (?)

Here is a screenshot of how it appears to me.

Screen Shot 2020-02-21 at 12 29 45 AM

created time in 2 months

issue commentarogozhnikov/hep_ml

weight normalisation

@jcob95, you should renormalize externally. As I understand your case, you should compute expected amount of samples in each bin first, and then within each bin you need to apply normalization so that total weight coincides with expected.

bifani

comment created time in 2 months

issue closedarogozhnikov/einops

pip package update? (for jax support)

Thank you!

closed time in 2 months

znah

created tagarogozhnikov/einops

tagv0.2

Deep learning operations reinvented (for pytorch, tensorflow, chainer, gluon and others)

created time in 2 months

release arogozhnikov/einops

v0.2

released time in 2 months

issue commentarogozhnikov/einops

pip package update? (for jax support)

Done. Try: pip install einops==0.2.0

znah

comment created time in 2 months

push eventarogozhnikov/einops

Alex Rogozhnikov

commit sha 00759ee327c4dec0cde3a0b93e50fe641c4dc48c

bump version

view details

push time in 2 months

push eventarogozhnikov/einops

Alex Rogozhnikov

commit sha 30017449eadb8f3bdd136e01d830adea55d70a83

minor comments

view details

push time in 2 months

push eventarogozhnikov/einops

Alex Rogozhnikov

commit sha 1788ebe812ef212bb792ff01d63bddd96ae76260

simplified error definition

view details

Alex Rogozhnikov

commit sha 830d3e6a86ff7debeca67a2177b73cc0993cb29f

update readme

view details

push time in 2 months

push eventarogozhnikov/arogozhnikov.github.io

Alex Rogozhnikov

commit sha 0f7d3efa2156885965b145be3703418a7480d9c9

new post

view details

push time in 2 months

push eventarogozhnikov/arogozhnikov.github.io

Alex Rogozhnikov

commit sha 25f3014c5e2cd8a01e687a6cc2246a6b7cbb0577

minor

view details

push time in 2 months

pull request commentstatgen/demuxlet

Add option to produce only singlets probabilities (fix #52)

@VincentGardeux thanks for the info, but I already wrote my own demultiplexer for rnaseq, I can close this PR if you're confident that problem with too many genotypes was addressed in your patch.

arogozhnikov

comment created time in 2 months

push eventarogozhnikov/python3_with_pleasure

Alex Rogozhnikov

commit sha aae5fa3b2916b8067746ed6ce515edf53783c42a

Update README.md

view details

push time in 3 months

push eventarogozhnikov/arogozhnikov.github.io

Alex Rogozhnikov

commit sha f6de5a0430bedae5a9d8017f7b6057cac83125bf

modified post a bit

view details

push time in 3 months

more