The filtered reads are aligned and mapped against the drosophila whole genome to identify the regions which are expressed in both young and old fly heads. The mapping is done using bowtie 2.0 to get the BAM file containing header, mapping quality and reads which are sequenced. There are 16 sets obtained containing reverse as well as forward reads aligned and a set of regions which are not aligned.

The aligned reads are obtained from BAM files generated by bowtie. These reads are further counted against the protein coding gene to obtain the annotation of reads. For this featurecount program is used to obtain the counts. The reads which are overlapped with the genes are counted and annotated. The number of overlapping reads also depends upon the depth of the sequence and quality of the sequencing. The count file is simple tabular file generated containing gene IDs and number of reads against them.

The obtained count file is used as input for differential analysis [11] [12] of the gene i.e. the reads which are counted against the gene is analysed for differential expression in young flies and old flies to find out the gene regulation in brain of old age flies compared to young flies. Limma voom [13] [14] package is used for finding differential expression of gene. The tool is used to analyse mean variance of the samples and normalisation of each sample count file. The mean variance analysis and statistical analysis is the measure of expression of gene in each sample which can further filtered to get the most differential expression of the gene [15] [16] [17]. The differentially expressed genes are established for gene ontology analysis to get the biological function of each gene and pathways in which these genes are involved. The pathways are also designed and deduced for identifying the actual connection of each gene with the other and hence gene regulatory network of the complete highly expressed genes is obtained.

Sequences obtained from different samples of aging male and female flies from SRA dataset are raw sequences containing noise and unwanted sequencing errors hence it needs to go through quality check to get efficient results, for this trimmomatic and cutadapt [18] program is used to remove noise and adapter sequences and it was found that around 60% reads are unique in reverse and 48.5% reads are unique in forward strand in paired end RNAseq data of aging flies’ heads. The refined and filtered data is deployed for mapping which is done by bowtie2.0 program against the reference genome of drosophila melanogaster to identify the genes which are affected in old flies in comparison to young flies. The mapping statistics is shown in (Figure 1) which indicates 40% - 60% of alignment which is found to be unique and round 20% - 40% reads are unaligned with the drosophila genome.

The BAM file is obtained as output file of mapping which is actually compressed form containing mapping information which is further inspected and visualised through IGV (integrated genome viewer) [19] [20] which shows mapped reads at each position and connecting lines between the aligned reads indicate reads mapped against introns (Figure 2).

For counting the number of reads many programs are there [21] [22] [23] corresponding to each annotated gene we used featurecount [15] program which takes the mapped BAM input and type of strandness as we have unstranded RNAseq data hence we got the counts in forward as well as reverse direction both. We also used a gtf. file as input to get the annotated genes against the reads. 20% - 30% of reads are mapped with genes and the numbers are found to be less due to use of a particular tissue type against the whole genome of flies (Figure 3).

To be able to identify differential gene expression [24] in old flies all dataset of total 8 samples having forward and reverse strands are analysed. 4 samples of young flies i.e. 5 days and 10 days old and 4 samples of old flies having 30 days and 40 days old are analysed and as we have already 8 files containing forward and reverse counts against each gene. Some samples have more reads compared to other samples which shows higher sequencing depths of the samples hence mapping of reads against the gene also depends on sequencing depth and length of the gene, longer the gene higher the reads found against them. So, to deal with such situation normalization of data is done and for that limmavoom [13] [14] is used to run Differential gene expression. The R script is used for limma package to normalise the count table obtained from different samples and this step is done to equalise the relative abundance of each gene in a RNA sample as in some cases small number of genes are highly expressed in one sample compare to another which may cause false positives hence normalisation is used for equalising the abundance. Here, we incorporated a factor called aging of flies comprising young flies and old flies. The data generated here contains the differential regulation of genes of old flies compared to young flies. And it was found that 141 genes are upregulated and 262 genes are down regulated in old flies as compared to young flies out of approx. 17,555 genes. The summary table is obtained containing gene identifiers, mean of normalised counts average to all samples, fold change in log2, standard error estimate for log2 fold change, wald statistics, p value, p value adjusted to multiple testing which controls FDR. A graphical summary of results is obtained which contains mean variance plot, mds and box plots and volcano plot (Figures 4-6) which show the variance and similarity in the gene expression of different samples.

This obtained data is filtered to extract and annotate differentially expressed genes by identifying absolute fold change greater than 2 and for that first the significant adjusted p value is calculated by filtering all values less than 0.05 and found 430 genes which are significantly expressed further this result is narrowed down by filtering abs log 2 fold value greater than 1. It was found that there are 350 genes which are expressed differentially with significant adjusted p value. The mean variance of table is calculated and Z scores are calculated and annotated using DAVID. The top 32 genes with Z scores are shown in heatmap (Figure 6).

The genes are further analysed to identify their presence in various biological pathways and to understand the process of aging in more depth. The Go analysis reduces the complexity by linking the gene directly to its biological and molecular function. Goseq program is used to identify the pathways related to these genes. The wallneius Rank category method is used to get corresponding GO

term over represented p value, under represented p value, number of differentially expressed genes, number of genes in this category and details of the term is calculated. A graph of top 10 over represented GO term is given in the (Figure 7) which shows that genes involved in stress response, apoptotic process, histone modification, covalent chromatin modification, protein modification by small protein and protein ubiquitination, immune and defence response regulation. The KEGG pathways plotted using pathview [25] against topmost expressed genes are also generated which shows complete information about the genes and their role in pathways involved in aging and dementia (Figure 8).

Our result provides the link between aging brain and dementia which can be further deployed to establish therapeutics against neurodegeneration and aging brain. As, the data shows that both run parallel to each other and factors affecting brain aging strongly contributes to neurodegeneration and dementia.