2. Why do we speak about AI?
• From ´60-s, programming… nice calling words. (first 1956 in a
meeting). Computer era has to come.
• Set-up of computing system is not ideal comparing with the only
intelligence, human brain. Calculation.
• Linear vs multi-level programming to copy and remodelling brain
networks (ie. remastering of probit-analysis to cover a wider spectrum of context).
Neuronal networking (copying Nature), deep machine learning, etc… a
great jump ahead as approach.
• Specialised processors, mixing of processor types – new era likely will
come.
4. Development: CPU GPU [A(G)I] 1.
• Visual information processing in brain ~70%.
• Not fully understanded, similar to parallel processing so called neuronal
networking: copied by TrueNorth, others on the field).
• Linear logistic and processing in classic CPUs is weak and slow pixel by pixel
processing - a great issue in normal world like self-driving cars. Timing?
Price?
• „GPU”s: improvement due to the parallel small classical computing units
and specific programming language. Achitecture specialisation. Possibility
of connecting units means higher capacity – likely data mining and
analysing. Nvidia – leading in business, 20,000 partner in 2017.
• TrueNorth: 1 million neuronal processing units, 4 cm2, low energy demand.
Fully replanned architectural system for mimicking neural processing of
information. Artificial vision is targeted.
6. The problem in genomics
• Workload and capacity to evaluate the info. Smart equipments, softwares,
databases to handle data. Repetitive tasks: sequencing (even full genomes)
or hot data of patients (smaller seqs, but numerous). Chip techniques.
• Milestone: Human Genome Project, 2004. One said it is a hammer. It
started to become from science to services – that was a change. Cost?
• Omics as examples where so called AI applications and programs were
helped a lot.
• Genetics. Genome data is a code. DNA. Genotype to phenotype as a gold
mine. Diseases?
• Bad genes and good genes. Risk analysis.
7. DDT Biosilico, 2005
Genome science first. Two simple methods are compered.
Proceedings from the IEEE, 2016
Deep machine learning models
Simple 1. Improved by ML 2.
Source of experimental data. Tissue specificity in 2.
8. Genomic regulation of information for biology:
genotype to phenotype (targeting diseases, personal medicines)
Costs and benefits
for companies
Insurance and healthcare
Enviroments?
Idealistic view a bit.
9. Life: How does genome works really?
Methylation
as epigenetic regulation
(cross-generation?)
microRNA and other
regulatory RNA forms
Post-translational modifications
like epigenetics enzymes on histones
Genomics structures
and pack of DNA
with proteins
Structural issues of
in/activity.
Additional facts (2016)
Environments
Complexity is well over than expected
Robotics, smart equipments, databases, curated by human workforce, intelligence.
Genome regulation is very complex.
10. Epigenetics: A Title from Frontiers series
• Frontiers in Toxicogenomics in the Twenty-First Century—the Grand
• Challenge: To Understand How the Genome and Epigenome Interact
• with the Toxic Environment at the Single-Cell, Whole-Organism, and
• Multi-Generational Level
Dec of 2017, Frontiers of Genetics
Remember the start: analytic approach. Here we speak about again analytic instead of holistic approaches.
Data and work overload is even bigger. CpG islands in genome. GWAS and EWAS: holistic approach: SNPs/methylation
sites for disease markers and risk analysis.
11. In a Middle of a Revolution: Enormous Results.
Everywhere in Science. Even in Healthcare.
• We can sit here untill midnight…
• Human genome. Neanderthals, Denisovians. Isolated population and
diseases in genetic trait. Mormons, Iceland.
• Human linkeages. Diseases and factors.
• Extreme genetic informations. Y and mitochondrial DNA. Microbiome of
digestive system in human.
• Those are positive things. Paleogenetics.
• Genetics spreaded though all over Science and became the part of it...
• My personal view is very positive on that. AI methods, if we can speak
about them currently, helped to spread it.