- Julia Wrapper for MINE
- website ltguo.me
- c++
- DSA
- first step with Julia
- first step with Julia
- reinstall ubuntu
- return key and card
- build julia, tensorflow, hexo
- first step with julia
- resign signments etc.
- shopping for basic life
- Jeff Dean's presentation on Deep Learning Summer School
- A tour through the tensorflow codebase
- move home
- build julia, openssl, libgit2, curl etc
- build julia, openssl, libgit2, curl etc
- learn from the 10th old competitor
- setup octopress blog system
- emacs various mode
- serizure prediction data preprocessing
- shared memory in different process
- write notes about lecuter 13,14 of cs231n
- write notes about lecture 8~13 of cs231n
- write notes about lecture 1~7 of cs231n
- Finish writing for company
- cs231n, cs224d slides
- tensorflow usage
- write a blog about bazel build in tensorflow
- add about in my blog
- basic c call cpp, julia call c demo
- Deep Reinforcement Learning one of Four without lab1
- MINE.jl
- learn emacs
- blog about emacs, gdb and c++
- get faimilar with tensorflow
- try to preprocess data with tensorflow
- learn the preprocessinng API of tensorflow
- build LSTM_CNN model
- the bug of pywrap_tensorflow disappeared and everything goes toward expected.
- learn various tutorials and sovle a segment fault (core dumped) due to out of memory
- learn how to build sophisticated RNNs and CNNs
- learn to work with tensorflow source code
- the layout of tensorflow source code
- swig
- understanding and visualizing CNN
- finished files for the development of NGS data analysis core software
- learn to how to build RNNs with TensorFlow
- fist step with Julia
- build simple LSTM to predict chr1 with torch
- build DBLSTM to predict sequence on chr1
- apply torch-rnn to Genome
- build Deep Bidirectional LSTMs with rnn in torch7
- LSTMs with torch
- prepare blog of dllab.org
- test post with octopress
- train CNN, LSTMs with Torch
- faimilar with LSTMs
- Genome data
- check the correctness of data and train benchmark model
- quick go through CS224d from 15:30 to 18:00
Refactor code of DeepSomaticSNV
- data preprocessing
- model various parameters
- various stats such as log analysis
- train sgRNA dataset
- two+ paper of sgRNA
- download ImageNet datasets
- train LeNet-5 CNN
- test ResNet on toy dataset
- STARS
- Mini Julia
- Julia for CUDA
- CodeAnalysis.jl for Julia, C, C++
- Kaleidoscope.jl
- COOL, COOL.jl
- new topology and opt algorithm for network
- baseline of Project S
- dig 14 paper
- data preprocess
- CNN/softmax as base model
- read 14 Broad's paper
- deep learning on genomics
- 14Broad's paper
- download all the genomes from ensembl
- vyper pipeline
- fix bug of keyerror
- MSR's code: not clear the way of hack
- GeneticAlgorithm called
- mechanisms of sv: look througth a little
- vyper pipeline and get breakpoint, not done
- weekly summary and plan
- DataExpore.jl
- MH question
- STRs and homopolymers
- sgRNA design basics
- GRE word 2 day
- JuliaCon machine code
- Julia.h julia_internal.h
- JuliaCon2016 part videos
- GRE 2 days words
- MH Vy-PER paper comprehension
- dive to Azimuth code and algorithm
- read paper of Azimuth
- try to learn some NoSQL databases
- collect data relevant to Proj S, the result is collect several labs
- understand 2016 but still a lot of questions
- read MR's paper again
- funetine pipeline of Vy-PER
- summary of last week
- plan of this week
- communicate with Hu
- reading paper from MR one time
- run vyper
- run azimuth
- finish vy-per testing
- reading all the code of azimuth
- rewrite final_fitering
- rewrite blatsam and test it on real data
- test julia version of Vy-PER and rewrite sam2fas
- test julia version of Vy-PER
- digest Vy-PER
- pass Vy-PER on next-generate sequence data without the final filter
- run Vy-PER on my data
- Vy-PER
- COOL
- WARNING: could not import Base.complement into DataStructures WARNING: could not import Base.complement! into DataStructures
- IO regression and readavailable(IOStream)
- fucntion sequence()::Int64 end not clear error msg
- methods of fieldname and fieldnames are not consistent
julia> type Test
a::Int64
b::String
end
t = Test(1,"hello")
Test(1,"hello")
julia> fieldnames(Test)
2-element Array{Symbol,1}:
:a
:b
julia> fieldnames(t)
2-element Array{Symbol,1}:
:a
:b
julia> fieldname(Test,1)
:a
julia> fieldname(Test,2)
:b
julia> fieldname(t,1)
ERROR: MethodError: no method matching fieldname(::Test, ::Int64)
Closest candidates are:
fieldname{T<:Tuple}(::Type{T<:Tuple}, ::Integer)
fieldname(::DataType, ::Integer)
in eval(::Module, ::Any) at ./boot.jl:225
in macro expansion at ./REPL.jl:92 [inlined]
in (::Base.REPL.##1#2{Base.REPL.REPLBackend})() at ./event.jl:46
- getfield exits but not getfields
- string concation * ^
- print_with_color is too long to write and use
- MH done
- HIVID make it done
- Bug: Pkg.update false
- Julia: julia.h, julia_internal.h
- HIVID: aln, overlap, merge, aln, location
- Finish Overlap.jl
- A sold understanding about julia.h
- make a reference contains hg19 and all hpvs
- HPVID base version
- HPVID base version [not finished]
- HPV integration
- MITx Unit3,4
- Fix BHTsne.jl
- Fix MINE.jl
- Familar with LLVM IR and Assembly
- LLVM
- Julia Deeper
- Assembly
- CPP
- LLVM
- ccall
StrPack.jl
to replacestruck
module in Python
- APIs of htslib
- htslib.jl bam APIs finished
- tensorflow
- learn how to use tensorflow
- pyjulia and cnn, lstm in tensorflow
- aliyu
- tensorflow, docker and aliyun
- cnn Mocha
- deep learning getting
- learn cnn
- make bam2pileup matrix to hdf5 file
- DarkVC
- papers about deep learning bio
- DeepSEA releated
- PPT
- compare prefermance of M2 and Varscan2
- machine learning applications on Gene
- compile gatk-engine
- deep learning on gene expression and predict varant effects
- DeepSEA
- learn Lua
- fix erros while building miniutils with bazel
- GATK
- Be able to use classes in GATK
- finish local assembly with De Bruijn Graph 1
- finish PairHMM 2-3
- finish FM-index aligner 4
- minigatk with bazel
- master Bazel
- minimum gatk with Bazel
- write a new walker based on Mutect2 with GATK in java and scala
- write a new walker based on Mutect2 with GATK in java and scala
- Fully understand intelij idea => part
- java unit test for GATK => GATK its own test
- how to write a new walker with GATK in java and scala => part
- De Bruijn Graph
- Test Mutect2 local assembly part with data
- how to use GATK in a new package both in java and scala
- Tests in GATK
- hack M2 throught
- Parallel MuTect2 on Cluster
- DSA TsingHuaX week 1 and rank 107
- nfs for x03 haplox success
- De Bruijn Graph for local assembly julia package
- PairHMM for pair alignment julia package
- spark standalone cluster success
- Presentation about M2
- make PPT
- pick algorithm
- deep understanding of M2
- parallel M2 on queue
- deep understanding of M2
- M2 Math part TODO
- M2 code part
- Parallel pipeline with Queue
- hack M2 throught
- learn scala and java
- learn spark
- learn java by Edx course
- learn scala by Courera course
- backgroud nosie mode
- pipeline parallel only GATK part
- MuTect2 Rewrite with scala with 3 days
- parallel MuTect2
- transfer pipeline to cluster
- go throught ScalabyExample
- scala cheatsheet and simple examples
- find feature for 6+9 samples of lung cancer and normal => gene mutations
- prepare clustering code
- hack Mutect2
- find feature for 6+9 samples of lung cancer and normal
- learn scala by 10miniutes
- learn java by 10minimutes
- hack Mutect2
- hack capp stanford
- write week report
- read papers and write varant callers review 50%
- Finish pipeline: finished 60%
- beta-bionomial distribution
- mpileup APIs of htslib
- CS231 assignment 1 50%
- download and browse most of variant call papers during 2012-2016
- finish most of code about pipeline and wait to bug
- Ng's ML about mixture of Gaussians
- Game Theory Week Two
- CS231n segmentation video
- Game Theory
- samples pileup stats
- Data scitentis toolbox at Coursera
- eval the feature of err baseline
- Simulate data with BAMSurgeon
- Variant Call with various callers
- Get underlining principle of BAMSurgeon
- VariantCall for ctDNA: check varscan2
- Fun.jl
- osx support
- stat tec support for production manager finished
- illumina barcode demutiplex finished
- osx support
- htsFile
- production related
- illumina barcode demutiplex
- tensorflow udacity assignments
- htslib
- 931 genes related
- htsFile
- htslib query
- fix htslib's bug
- fix bam query bug
- prepare notebook etc for home
- bam/sam hdr write
- bam index
- bam write and bam header IO
- hight level wrapper of bam read, write and query
- bam read and bam write
- bam read formal form
- paper about tumor evolution
- DL of Google
- L1,L2,L3 of DL course from Google
- Deep Learning from Google
- make read bam true
- SeqErrorDetector draft
- HTSLIB.jl
- looking for extreme sparse models in ICML NIPS ICCV and other top journals
- vcf data mining
- assign 1
- week summary
- lecture 3
- check models
- hand compare mutations between Small intestine and Biliary tract
- predict
- model vcf part finish
- fix codes about work
- lecture 2 of cs341n
- check basset
- a good materal: cs341n:Convolutional Neural Networks for Visual Recognition
- Finish work not done of this week
- Apply Deep Learning to kaggle tasks
- report
- MXNet.jl
- apply MXNet.jl
- rewrite Fusion dectect codes
- use all the haplox data
find the methods of handle false positive
finish the model part
finish Strict except a good model
hackerrank python
9:30-11:00 readdoc,julia
11:30-2:00 hackerrank python
2:00-5:30 apply tensorflow to one kaggle dataset
5:00-7:00 spark,scala
8:00-22:00 lan,network,htslib
readdoc and intro to network, hackerrank, julia, lan
one or two machine learning or deep learning package
spark, scala
learn Readdoc and hackrank
learn introduction to network of stanford
just remains a few bugs of work. Go to Deep for now.
focues the classification task.
make GeneMisc ready for users and prepared to regeister
test my model in real cfDNA data
finish and test gene_location and gene synonym
Summary about my skills:
One, mathematics including machine learning, mathematical optimization, PGM.
Two, computer science including DSA, GPU computing, compiler.
Flash extended and Alogorithms_stanford urgent
finish Flash Project
too much ideas to execute so that achived nothing
First, classification with an emphase on feature selection
Second, GP tune hyperpara
Third, PGM / DBM
1. Learn the structure of COSMIC by PGM
2. feature selection with random search
A great idea come in mind.
chose hyperparameter and do feature selection with Guassian Process
summary current published work on cancer risk prediction
Plan daily since 10/26/2015